86% of Windows 7 PCs Maxing Out Memory

CWmike writes "Citing data from Devil Mountain Software's community-based Exo.performance.network (XPnet), Craig Barth, the company's chief technology officer, said that new metrics reveal an unsettling trend. On average, 86% of Windows 7 machines in the XPnet pool are regularly consuming 90%-95% of their available RAM, resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks. The 86% mark for Windows 7 is more than twice the average number of Windows XP machines that run at the memory 'saturation' point, and this comes despite more RAM being available on most Windows 7 machines. 'This is alarming,' Barth said of Windows 7 machines' resource consumption. 'For the OS to be pushing the hardware limits this quickly is amazing. Windows 7 is not the lean, mean version of Vista that you may think it is.'"

Comments

When do people get this

[sopssa]

RAM is wasted when it isn't in use. The fact that the task manager in Windows says your RAM is used 95% tells nothing, and no it won't "result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks". I'm actually really surprised, and not in a good way, that "chief technology officer" of the company doesn't know this.

The new memory models in recent OS's try to utilize all the available RAM (as they should) to speed up things otherwise. It makes a lot of sense to cache things from hard-drive in low-peak usage points, and in such such way that it doesn't interfere with other perfomance. When the things that are most often used are already cached in RAM, their loading works a lot faster. This doesn't include only files, icons or such, but everything the OS could use or do that takes time.

If theres a sudden need for more RAM, the cached data can be "dropped" in no time. It doesn't matter if it averages at 25% or 95%, just that the perfomance overally is better when you utilize all the resources you can to speed up things in general.

Re:When do people get this

[Mr Thinly Sliced]

My understanding was that memory used for disk caching doesn't show up in task manager as "used".

It's been a while since I booted win7 though, so I might be mistaken.

Certainly under linux ram used as disk cache is marked "free".

It wouldn't surprise me that win7 has a heavier memory footprint though - as more applications move to .net and web browsers use lots of flash / silverlight etc - all of these things have a RAM cost.

Re:When do people get this

[Qlither]

Great post, if i had the points i would mod you now.
As i happens mine does not use much over 75%-85% when watching videos in firefox or playing games. It is far more effective if programs load all the common use files into ram like firefox.

Re:When do people get this

[dr.newton]

If all that RAM was simply being used for a filesystem cache, the system would not have to "increasingly turn to disk-based virtual memory to handle tasks" - it would just drop some cache when it needed to start a new task, as you said.

It seems that something else is going on.

Re:When do people get this

[Bazer]

I don't know how caching works in W7 but on Linux, if the system has to "turn to disk-based virtual memory to handle tasks" then the memory utilization isn't caused by buffers because buffers are never swapped out to disk. If W7 behaves in a similar manner the it's either a memory leak, system bloat or the caching mechanism is broken.

Re:When do people get this

[eldavojohn]

The new memory models in recent OS's try to utilize all the available RAM (as they should) to speed up things otherwise. It makes a lot of sense to cache things from hard-drive in low-peak usage points, and in such such way that it doesn't interfere with other perfomance. When the things that are most often used are already cached in RAM, their loading works a lot faster. This doesn't include only files, icons or such, but everything the OS could use or do that takes time.

If theres a sudden need for more RAM, the cached data can be "dropped" in no time. It doesn't matter if it averages at 25% or 95%, just that the perfomance overally is better when you utilize all the resources you can to speed up things in general.

Assuming your claims of how Windows 7 is implemented are true, then the claims from the person who actually collected all the empirical evidence must be false:

resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks.

If the memory was freed up dynamically as needed then no processes would ever be forced to resort to disk-based virtual memory. So either you work at Microsoft and are assuring us that the implementation protects against this or you're speculating against someone who has claimed to gathered a large enough to make such accusations.

No offense but I'm going to side with the guy who appears to make his living testing these sorts of things ... the guy who is offering me numbers.

Re:When do people get this

[ilovecheese]

Although I don't do any gaming on my Windows machine here at the office (Win7 x64), I rarely exceed 50% usage of my installed RAM (4GB). Running a lot of ssh terminals, browser, email & skype...

Re:When do people get this

[Rockoon]

According to their website blog, they are "rethinking windows performance."

So instead of thinking about what actually effects OS performance, they are rethinking things so that they don't have to sell real solutions to their customers, "where [they] maintain several large installations of our commercial DMS Clarity Suite performance analysis solution."

Re:When do people get this

[FlyingBishop]

Bull. Active RAM chews energy, which is terrible for a portable. It also causes problems for the (fast-disappearing) desktop, which will heat up when that much RAM is used.

And it will in fact cause a slowdown when the OS has to free up that RAM cache to load a large file from disk. Frugality is a virtue.

Re:When do people get this

[democomputer]

The article doesn't state that they have measured increase in virtual memory usage, just that it is a consequence of running out of RAM.

Re:When do people get this

[Anonymous Coward]

I think the issue here is that the system is turning to swap. Caching stuff that may be referenced again is fine and dandy, but if the system regularly turns to swap just to keep itself afloat, then you have a problem.

Re:When do people get this

[Tomsk70]

I'm sure the disk thrashing is a problem too, and not really the indexing service doing its' thing.

This sort of thing is what happens if you don't actually use the software you're criticising...

Re:When do people get this

[biryokumaru]

Um, my three year old WinXP system doesn't go over 15% when watching 720p video or playing FPSes. Are you certain there isn't something horribly wrong with your computer?

Re:When do people get this

[dr.newton]

From TFA:

"On average, 86% of Windows 7 machines in the XPnet pool are regularly consuming 90%-95% of their available RAM, resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks."

Re:When do people get this

[jibjibjib]

Using more RAM doesn't use more energy. Either your RAM is powered on, or it's not. And if it's powered on it maintains its contents, no matter whether the OS has actually written anything useful to it.

Re:When do people get this

[Anonymous Coward]

RAM is wasted when it isn't in use. The fact that the task manager in Windows says your RAM is used 95% tells nothing, and no it won't "result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks". I'm actually really surprised, and not in a good way, that "chief technology officer" of the company doesn't know this.

honestly, when have you ever seen a CIO or CTO that knows anything about technology? most of them are so out of touch that they still think that all of the computers and servers are plugged into hubs instead of switches. imho, most CIO/CTOs get there because of who they know, not what they know.

Re:When do people get this

[jimbolauski]

RAM is wasted when it isn't in use. The fact that the task manager in Windows says your RAM is used 95% tells nothing, and no it won't "result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks".

While there are approaches using virtual memory that will limit the slow downs, the data still has to come form the HD even SSDs are not faster then ram. Grandma checking her email may not notice but data processing and other memory intensive functions need memory and putting in on a HD will slow things down.

Re:When do people get this

[Sockatume]

If they'd measured pagefaults, they could've reported pagefaults. They didn't. RAM usage appears to be the total basis for the article, so his concern is a genuine one. We don't know enough about the study at this stage to dismiss it.

Re:When do people get this

[snemarch]

It shows up as part of the memory commit bar - which is what regular users will look at, and then go off screaming about "OMG IT USES ALL MY SYSTEM MEMORY!1!!! one one". It's also deducted from the "free" count, since technically it isn't free (it can be freed quickly, but has to be zeroed out before it can be handed off to a new app - security and all).

The Win7 task manager does show a "cached" stat, though, so your effectively free memory is "free"+"cached". And if you want more comprehensive memory stats, you should look at perfmon.msc or SysInternals' Process Explorer.

I wonder if TFA has actually measured that disk swapping happens (easy with procexp or perfmon), or are just shouting their heads off without understanding what's going on... it's well-known that SuperFetch utilizes otherwise unused memory for disk caching, and does so proactively :)

Re:When do people get this

[Anonymous Coward]

>.> See, they monitor metrics like this, BUT they have no way of actually measuring what the user sees. And the "resulting in slow-downs" is amusingly unsubstantiated.

Re:When do people get this

[jimicus]

Here we see why /. needs a "-1, Wrong" mod.

Re:When do people get this

[biryokumaru]

Oh, right, I can read. RAM.

Disregard the above. I'm dumb. =[

Re:When do people get this

[Anonymous Coward]

I don't know about you, but I run W7 and I don't know about these "maxing out" 's, my computer seems to run as fast as it did when I had XP (which is actually surprising).
Maybe the man DOES have empirical data, and maybe he does work with numbers, but what I think matters the most to a user, is his/her own experience. And in my experience, my W7 runs very well, consuming the same amount of RAM as XP did (and definitely less than Vista).

Re:When do people get this

[slim]

Yes. I have a similar problem when people running servers complain that the CPU is at 100%.

If you're seeing an actual slowdown in performance, fine, worry about it.

Otherwise, 100% CPU usage is a good thing: it means there's a process that's not IO bound.

Re:When do people get this

[ThaReetLad]

Reads like guess work to me. This is statistical data collected by an automated tool. If this tool were ACTUALLY measuring and reporting increased swapping, then they would report that, but they don't.

I call fail on Craig Barth.

Re:When do people get this

[dskzero]

I'm not an expert of memory management in OS, but if the RAM is being used, well, it doesn't means there is an impact on the OS. The RAM is being used for *something*, that's it.

Re:When do people get this

[Mr Thinly Sliced]

Ahh fair enough. "Colour me learned something today". :-)

Re:When do people get this

[mister_playboy]

The new memory models in recent OS's try to utilize all the available RAM (as they should) to speed up things otherwise.

BeOS and Haiku did/do this, but I don't think any other OS has implemented total RAM usage to such a degree.

Re:When do people get this

[phatcabbage]

Barth acknowledged that XPnet's data couldn't determine whether the memory usage was by the operating system itself, or an increased number of applications.
So yeah, it doesn't seem like the author really knows what's going on...

Re:When do people get this

[dhavleak]

If the memory was freed up dynamically as needed then no processes would ever be forced to resort to disk-based virtual memory.

The trouble is, the TFA doesn't actually say (at least not clearly) that the Win7 machines are indeed turning to swap more regularly. It just states that fetching stuff from the swap file is a consequence of running out of RAM and causes perf degradation. So if the Win7 machines are indeed utilizing all available RAM and yet not swapping at a significanly higher rate, it means they're making more optimum use of available RAM.

Re:When do people get this

[ThaReetLad]

Not on slowdowns he isn't. The data is memory usage, but the system slowdowns claim is supposition from the first number. He says the memory usage number implies slowdowns, but that conclusion is false on win 7.

Re:When do people get this

[Anonymous Coward]

Bull. There is no such thing as "Active RAM". 0's and 1's use exactly the same amount of energy.
In theory you could disable the refresh cycles for unused portions of the RAM chip, but because the RAM chip decides which part to refresh, this will be very difficult.

Bull(2). The OS does not spend any time to "free up that RAM cache". the RAM is simply overwritten with new data, just like if it was free. The original data was just a copy of data from the HDD.

Re:When do people get this

[snemarch]

TFA (which I've just read :)) doesn't really mention whether they're looking at disk paging. Browsing to the xpnet site does say they do, but it also mentions that they base memory load on "free physical memory". The Windows APIs don't report the superfetch cache memory as "free physical memory", since it isn't.

Those Devil Mountain guys need to update the system metrics they're collecting - it should include at least cache-size in addition to free physical memory, and pagefile use (size of paged-out data and paging rate per second!) should be a separate stat, not factored in with the "free physical memory" stat.

No offense but I'm going to side with the guy who appears to make his living testing these sorts of things ... the guy who is offering me numbers.

He's offering you interpreted numbers, not raw data - and he's not offering you all the numbers you need in order to make educated statements on what's going on. He's probably doing it in good faith out of ignorance, rather than trying to spread FUD, but it's still pretty useless metrics.

Re:When do people get this

[dunezone]

Hi there.

Ive been running Windows 7 since the BETA release. I have never experienced any issues that result in I/O thrashing against the hard drive as result of all my ram being utilized. I also have numerous friends running Windows 7 none have reported any issues like this, if anything its been praise for the operating system. So I am amazed to see a 86% number being thrown out there yet never seen this problem before.

No offense but I'm going to side with the guy who appears to make his living testing these sorts of things ... the guy who is offering me numbers.

And now lets quote something from the article...

Barth acknowledged that XPnet's data couldn't determine whether the memory usage was by the operating system itself, or an increased number of applications, but said that Devil Mountain would start working on finding which is the dominant factor in increased memory use.

This single sentence makes the article rubbish. They have no clue whats causing the heavy memory usage, its just an assumption that the OS is causing it and they're yelling fire before looking through all the data or completely analyzing the problem.

Re:When do people get this

[ysth]

I wonder if TFA has actually measured that disk swapping happens (easy with procexp or perfmon), or are just shouting their heads off without understanding what's going on...

I give you one guess.

Re:When do people get this

[nmg196]

Totally agree. If you don't want Windows 7 to use the 4GB of RAM you've paid for to speed up your computer, take out 2GB and put it in the drawer. Otherwise, be thankful that it's actually making the most of the RAM you're using.

What next? People complaining that games use 100% CPU to give them maximum framerate when it could just use 30% CPU and give them 10 FPS?

Re:When do people get this

[theaveng]

Or people that take the time to explain WHY my post was wrong. Which part is wrong? Where I said my brothers Win7 machine runs slower than my XP machine? That's not something you can demonstrate, because you have not seen his machine or my machine.

What else was wrong? The part where I said Puppy Linux runs like a speed demon, because it sits wholly-and-completely in RAM? Perhaps I did make an error there, but I don't think so. Puppy IS fast.

Maybe the part about HDD caching slowing things down?

I could be wrong there, since I'm not an expert but I remember the dark, dark days when my computer when spend 2-3 minutes just to redraw a Word document. Why? Because it was using the HDD like memory, instead of using the actual memory. It seems to me that this problem, while minimized, has never completely gone away.

Anyway telling me "you're wrong" doesn't enlighten either me, or the other readers. Please elucidate.

Re:When do people get this

[Josh04]

Someday I'd like to have a system where I have enough RAM to turn-off HDD caching completely.

The only system where it makes sense to disable swap space is a system with no HDD at all. Why would you prevent your operating system for accounting for the possibility of filling all that RAM? Do you really think the OS is trying it's hardest to make your computer go slower?

Re:When do people get this

[kangsterizer]

oh, i suddenly get why most people are uninformed. time to look like we're all super compworld pros with numbers (doesn't matter what content we put,we're trusted!)

Re:When do people get this

[snemarch]

My guess would be "nothing to see here people, move along - trolls day out, looking for publicity".

Re:When do people get this

[Anonymous Coward]

RAM is wasted when it isn't in use. The fact that the task manager in Windows says your RAM is used 95% tells nothing, and no it won't "result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks". I'm actually really surprised, and not in a good way, that "chief technology officer" of the company doesn't know this.

You nearly nailed it.

CTO's are more concerned about getting the best performance out of their tightening budgets. Windows reporting 95% ram utilization gives them no visibility of real world performance apart from when there users get annoyed enough to start complaining about how slow it is.

Problem with that they may not have the budget anymore to speed up network, or upgrade machines etc. This may not even be driven by the operating system itself. for example Say a new CRM platform requireing more local machine resources is installed. Windows will still report av of 95% utilization but the real world performance might be completely different.

Re:When do people get this

[FlyingBishop]

You'll excuse my ignorance, but from college I remember that usually you have 0-2V represent 0 and 3-5V represent 1. Does a 0 have a corresponding increase in amperage so that it levels out and uses the same amount of power?

It seems natural to me that it would be initialized with zeroes on power-up, so that it would minimize power consumption.

Furthermore, more advanced chips, especially in mobile devices, have a variety of power-saving tricks. I would expect RAM would be no exception in having ways to clock down in line with requirements.

Re:When do people get this

[Fantom42]

RAM is wasted when it isn't in use.

You make decent arguments that I've heard before, many times. And I actually agree with them to a point. At least in theory, given somewhat unrealistic assumptions about the system. But there are two major flaws in this idea. First, in practice, where real developers are making complex decisions about how they structure their programs and the resources they use, I think this sentiment results in wasteful and excessive programming practices.

In other words, the resources programs demand may be managed very well, but programmers end up demanding more than they need, forgetting that there are other users of those resources. And a scheduler isn't psychic; it has no real way to know if a program is demanding "too much."

Second, a multiqueue system with a nonzero variance of demand (e.g., a computer's resources) has high latency if its resource utilization is also high. This is due to bottlenecks being created when there is a surge in demand, and resulting system instability as those backups are cleared. If you want to have a system with consistent and predictable latency, you need some extra resource hanging around to handle spikes in demand. Otherwise the system tends to go berserk in a way that strongly resembles "thrashing" seen on computers that are being maximally utilized. People forget (or don't know) this.

Re:When do people get this

[Anonymous Coward]

If the memory was freed up dynamically as needed then no processes would ever be forced to resort to disk-based virtual memory.

That's just nonsense. Assuming you're not referring to the case where overcommitted memory is actually being used, here are other scenarios where paging to disk is desirable. For example, if a process is not being used (but is loaded) then it's good if its pages are paged out to make better use of physical RAM. Applications are generally greedy and unruly kids. They will ask for as much memory as they see available even if they never use it. The memory manager is tasked with keeping the kids happy (fulfilling their memory requests) while ensuring that all the kids can play together.

But based on the false dichotomies in your post, I assume you're either trolling or have no idea how memory managers work.

Re:When do people get this

[jernejk]

I used Ubuntu for almost a year and I think linux cachnig / virtual memory is implemented better than win7. It seems win7 cache is too aggressive and it dumps active programs from RAM to page files when it should not. Maybe it works OK for most desktop users, but it doesn't work very well for a development machine. I have nothing but my subjective feeling to backup my observations.

Re:When do people get this

[snemarch]

Well, it's easy to get wrong - and considering Windows is used by regular Joes, it's OK that task manager simplifies matters a bit instead of showing up pages of detailed information.

It's a shame when people who have no clue about how the memory manager works then start making a lot of assumptions... and even worse when somebody who's "collecting system metrics" don't seem to get it, either.

I'm not ruling out that there could be disk paging involved, and there's some actual problem, but considering my own win7 experience across a number of machines as well as the info in the TFA + linked site, I somehow doubt it :)

Re:When do people get this

[slim]

The only system where it makes sense to disable swap space is a system with no HDD at all.

Or an OS with terrible swap algorithms.

Anecdotal, subjective and unscientific: I perceived an improvement in performance when I disabled swap in XP.

Re:When do people get this

[theaveng]

I do energy analysis for systems, and "active RAM" does in fact use more energy than idle RAM. Not a huge difference but enough to add-up over multiple banks and affect the size of a power supply (for worst case). Active RAM has the same periodic refresh cycle of idle RAM, but also the constant reading/writing of data over the system bus, which means about twice the energy used.

Therefore I concur with the grandparent poster's statement.

Re:When do people get this

[ijakings]

Stop having such irrational thoughts.

Re:When do people get this

[Mr Thinly Sliced]

If the RAM is filled with a lot of crap, such that the OS must keep loading of the HDD to get information, that slooooows system performance not speed it up.

Actually in modern operating systems RAM can be used as a disk cache is such a way as it can be "freed" at little to no cost when needed by programs. This in effect means you get quick access to often used programs as they are already cached in RAM.

The only situation in which "RAM is filled with crap and the OS goes to disk" (paraphrased) is when RAM is full - with programs. In this situation yes, it has to go to disk.

Re:When do people get this

[SharpFang]

Windows is notoriously reluctant to invalidate caches to free RAM for applications. I don't know if Win7 fixed that, but XP would much rather send an idle app to swap than free some disk buffers. That's why switching swap off entirely tended to speed it up so much - it was forced not to swap out any active data and free up buffers instead.

Re:When do people get this

[dhavleak]

"-1 Wrong" on so many levels.

The reason puppy linux fits in 30MB and runs so fast, is because it does barely anything at all. Whatever purpose you use it for, it surely achieves that well, but don't make silly comparisons.

Next, the swap-file isn't "HDD caching". A cache hierarchy works the other way around -- the fastest (and consequently most expensive, and thus smallest) memory types are closest to the processor (L1, followed by L2, often followed by L3), and the slowest (and consequently least expensive, and largest capacity) are further away -- that would be RAM, and then the HDD. Why add the HDD to the hierarchy? 'cos that's what enables you to simulate infinite memory (or enough for the task at hand, even if you exceed your physical RAM -- if you have enough for the task(s) at hand, it doesn't matter that your capacity is actually finite). Long story short -- in theory, you could call just about every form of memory in your computer a cache, except the HDD.

Your 'anecdote' about your brother's Win7 / Vista (were you being snarky, or does your brother dual-boot?) machine, is what is commonly known as "a lie". Win7/3GB RAM/AMD X2 running slower than XP/512MB/P4? Sorry dude -- simply doesn't add up.

And finally, yes, retreiving stuff from the swap file too often kills performance. But who said that the RAM was 'full of crap' on Win7?

Re:When do people get this

[snemarch]

Yep, that would be a problem - but neither the TFA nor xpnet mentions if this is actually happening, it seems that they're looking almost exclusively at "free physical memory", which isn't a useful stat in this regard. The xpnet site does say they factor in "how often it relies on virtual memory", but not how they do this (there's multiple metrics to choose from, some fairly uninteresting) and the fact that they seem to factor this in as a part of "memory usage" rather than keeping it as a separate stat makes me pretty wary of trusting any analysis from them.

Re:When do people get this

[h4rm0ny]

While your point may be valid, I have a system without swap but with a hard drive (three actually). Why? Because I never want it to use swap. Ever. I loathe the idea that my system should resort to that. If I ever find it suffering due to lack of memory, I shall buy more. Elininating swap is done on aesthetic grounds. I renounce it entirely!

Re:When do people get this

[realityimpaired]

I had a great reply all typed up but the stupid filter thinks it used too many "junk" characters... not entirely sure what those junk characters are, but meh.

I'm on a 2-year old laptop with 4GB of RAM, which I use for gaming. Said system is running Windows 7 x64. Looking at the memory tab, there's two numbers that guage the available memory... of they're only looking at the one labelled "Free", they're going to see that I'm using 80% of my available memory. If they look at the one labelled "Available", they're going to see that I'm using 20% of my available memory right now. That's because the system is pre-caching more than 2GB of data into physical memory so that I don't have to wait for it to be read off the hard drive when I open a program. And it does make a noticeable difference... Office, Acrobat, games like WoW and DA:O all open noticeably faster than they did under XP Professional on the same system. Anecdotally, of course. But when playing games on that system, I *never* run into cache usage. In fact, out of the box Win 7 tries to avoid using swapping, and I'm currently sitting at zero swap usage.

In Linux, as you say, the "free" command simply flags cached memory as available. The breakdown of memory that is currently cached and can easily be dumped to free up memory simply isn't there unless you look at the +/- swap buffers line.

As others have said: precaching stuff into RAM is a good thing. It reduces lag when you're opening things, and can easily be dumped to free up the memory if you open something that needs the memory. I suspect that the authors of TFA aren't looking at how much memory is flagged as available, but rather how much physical memory is actually listed as free.

Re:When do people get this

[Eskarel]

There is no speed difference between loading data off the hard drive into empty memory vs loading the data into memory which is being used for cache. You still have to load the same memory either way, and wait for the same delays.

The reason you get memory issues when your memory is over full is because the data which is being taken out of memory may either be modified(in which case it needs to be written back to the disk cache) or reread at a later date.

If the data doesn't need to be written back to the system(because it hasn't been modified) and isn't going to have to be fetched again in a few seconds, then you get no performance degradation at all.

Generally speaking you will get much better performance having RAM full of stuff you might need, as opposed to empty since if you do need it you won't need to fetch data from the disk.

Re:When do people get this

[djnforce9]

This is exactly what I was thinking when I read that article. That CTO must be quite ignorant of the new model (I think even Windows Vista did the same thing if I recall correctly). Hopefully someone brings him up to speed on this.

I must say though that this new model does make it very difficult to get an accurate reading of how much memory is actually being used since the task manager does not differentiate between what has been reserved as cache and what memory space is actually occupied by applications. The best you can do is look at your list of running processes and make an estimate. Some of them may actually be in the pagefile though even when there's free RAM available because they are background applications that are used very rarely (or haven't been touched in an extended period of time).

I'm sure there are third party applications available though if you need a dead on accurate reading.

Re:When do people get this

[AliasMarlowe]

the TFA

At least you didn't write "the TFA article"

Re:When do people get this

[Bert64]

The article mentions that they use a program of their own creation for monitoring memory usage, so how the users interpret the data is irrelevant as the program will send what it believes to be correct.
Wether the program is accurate or not is another matter, but the fact it doesn't report every system as using 100% of its memory suggests it is at least somewhat aware of superfetch etc.

Re:When do people get this

[clone53421]

What next? People complaining that games use 100% CPU to give them maximum framerate when it could just use 30% CPU and give them 10 FPS?

That’s not a fair analogy. When you are playing a game, you are not multi-tasking. You are basically using the computer for a single task. I don’t care if the game sucks up 100% of the CPU and all of the remaining memory, as long as it frees it up when I close it.

If the system is gratuitously using 95% of the RAM nearly all the time, then it’s a completely different scenario. Everything I try to open that wasn’t cached already will force the system to dump some memory to the swap file to make room for the new application.

Re:When do people get this

[tgd]

Frightening, huh? CTO and, in fact, he has no idea what he's talking about.

Rather than blindly following presumed authority, perhaps you should make use of some critical thinking skills and a couple Google searches. You'll (apparently) be surprised what you learn.

Re:When do people get this

[Eskarel]

It depends what they're monitoring.

Looking at my windows 7 system. it does appear that you are correct and disk cache is not included in the "In Use" Category. However it isn't included in the "Free" category either, but rather in "Available", so if they're crappy software is reading the "free" category instead of "available" that won't include cached disk.

It's also possible of course that they're monitoring software leaks like a sieve on 86% of windows machines, which is entirely plausible.

Re:When do people get this

[snemarch]

You're making an apples-to-oranges comparison, and you don't mention what software is running on either of your machines...

I've found Vista and Win7 to generally work better, giving decent enough hardware. Because of SuperFetch, Visual Studio loads faster on my Vista64 dualcore laptop with 7200rpm drive and 2 gigs of memory than on XP64 quadcore workstation with 10000rpm raptor drive and 8 gigs of memory. But this isn't a fair comparison either, since the rest of the software suite on those two machines are specced differently.

Re:When do people get this

[lcarnevale]

HDD caching and swap are two completely different things. HDD caching is loading things from the disk to RAM to speed up things. SWAP is using the HDD as extra RAM when the system doesn't have any more memory left to use. So, what I think you wanted to do was to turn off swap, not hdd caching.

Re:When do people get this

[FooBarWidget]

Definitely the part about HDD caching slowing things down. Even in the DOS age it was well known that hdd caching utilities (I forgot the names, too long ago) improve disk performance tremendously.

Linux does the same things as Windows: it caches as much stuff from disk into main memory as possible. Try running:

    cat large_video_file.avi > /dev/null

You'll see that after running the command, your memory usage jumps up by the size of the video file. Now try running the same command again, it's now an order of a magnitude faster.

On Linux things like this are stored in main memory in the form of caches and buffers. I don't know about Windows, but Linux clears some caches and buffers if applications need real memory. Caches and buffers show up in memory usage reporting tools like 'free', so it's quite normal to see Linux systems using 90% or more RAM, most of which go to caches and buffers. It seems that most people who complain about memory usage don't know how memory is managed on modern operating systems, so they go all apeshit about "OMG HELP linux is using so much memory it sux0rz!!!" and I have to explain again and again how they're not getting it. Same goes to you. Now, Windows is suffering from the same problem.

FYI, here's the memory usage of my Linux server:

             total       used       free     shared    buffers     cached
Mem:           720        702         17          0         55        510
-/+ buffers/cache:        136        583
Swap:          399          0        399

It says 702 MB of used memory. Now look at "-/+ buffers/cache", it says 136 MB. That's the amount of memory *actually* used by applications.

Re:When do people get this

[cgenman]

Read TFA. It just claims that Windows 7 consumes all available RAM. That is the "empirical evidence." System slowdown was NOT measured.

Utilizing all available RAM is a pretty well understood technique at this point. All web browsers do this now, as do many other applications. One would expect a well-designed modern OS to do this. Consuming all memory itself is not a sign of poor programming itself, so long as disk caching of things that should be in RAM doesn't occur. This is not something that the people in the article has measured.

I'm going to side with the guy who appears to make his living testing these sorts of things.

Bad science is bad irrespective of the person conducting it. And whatever the original tester said is getting filtered through the viewpoint of the gentleman writing the article. Considering that he says that "Windows 7 is not the lean, mean version of Vista that you may think it is," yet never once compares statistics to Vista (or even mentions it outside of this statement), I'd take the conclusions from these stats with a grain of salt.

Re:When do people get this

[Barny]

One point, turning off the page file is never a good idea, it means that the few things that can be paged out quite happily without negative effect stay in memory, taking up room the cache could use to make the rig run faster.

My current machine has gone from XP to vista to 7, and I gotta say 7 is the smoothest, vista was the most stable, and XP, well, it "seemed" fast, but that was because it tied up one of my CPU cores all the time waiting to draw the next window.

Your brothers machine could do with a nice modern vid card or some cleaning up by the sound of it, on a modern PC 7 interface shouldn't lag.

While yes, puppy linux may run super fast, its also lacking a lot of what makes people stay with windows ;)

Re:When do people get this

[Moryath]

You obviously don't understand memory access design. It's all about feeding the CPU. There are two sorts of relationships we can use to make this work: temporal and sequential.

Hard drives are the largest-capacity storage (well unless you want to go to tape). But they're slow. Even the fastest high-RPM SCSI or SATA drives are SLOW compared to what's above them. This is mitigated, somewhat, by putting some cache memory on the drive's controller board itself. Still, having to "hit" the hard drive for information is, as you say, a slowdown. Same goes for "external" storage (Optical media, USB media, etc).

So you try to keep as much information as possible in RAM (next step up). Hitting RAM is less expensive than hitting the H/D in terms of a performance hit. In the original days of computing (up until the 486DX line for Intel CPUs), RAM and CPU operated on a 1:1 clock speed match, so that was that.

Once you factor in the "clock multiplier" of later CPU's, even the fastest RAM available today can't keep from "starving" the CPU. So we add in cache - L3, L2, and L1. the 486 implemented 8KB (yeah a whole 8K, wow!) in order to keep itself from starving. L3 is the "slowest", but largest, L2 is faster still but smaller, and L1's the smallest of all, but the fastest because it is literally on the same die as the CPU. That distinction is important, and in general you'll find that a "slower" CPU with more L1 Cache will benchmark better than a "faster" CPU with less.

The CPU looks for what it wants as follows:
- I want something. Is it in L1? Nope.
- Is it in L2? Nope.
- Is it in L3? Nope.
- Is it in RAM? Nope.
- Is it in the H/D Cache? (helps avoid spin-up and seek times) Nope.
- Crap, it's on the H/D. Big performance hit.

Everything except for the L1 check, technically, was a performance it. The reason for pre-caching things (based on temporal and sequential relationships) is all about predicting and getting what will be needed next into the fastest available place.

Yes, I suppose you can run an entire system where it all goes into "RAM", and you'll see it as "more responsive" simply because you never have to touch the hard drive. But turning off HDD caching is a BAD idea. It makes cache misses that much more expensive because then, instead of having even the chance of finding what you needed in RAM or in the HD's onboard cache, you have to wait for the H/D to spin up and seek to the right sector.

Re:When do people get this

[snemarch]

Also anecdotal, subjective and unscientific from my side: Windows seems to page out to disk too early, at least the "older" (and memory-conservative) releases, WinXP and below. After upgrading to 1GB of RAM and disabling pagefile, I definitely noticed less disk activity... and since then, I never looked back.

While the GP is correct in the dangers of disabling pagefile, I rarely had problems on 1GB, and never on 2GB. These days my machine has 8GB, so whatever :)

Re:When do people get this

[theaveng]

>>>The only system where it makes sense to disable swap space is a system with no HDD at all.

Or a system where, when you click on an icon, it opens the windows *instantly* not 1-2 seconds later (because it's retrieving data off the HDD). If you don't know what I'm talking about I respectfully suggest going to here and trying this OS on a bootable CD. You'll be amazed. http://puppylinux.com/

A computer that uses no hard drive caching, and runs completely in RAM, is faaaaaast. Of course with Windows, that means you'd need somewhere around 20 gigabytes of memory. Windows won't work otherwise.

>>>Do you really think the OS is trying it's hardest to make your computer go slower?

Apparently you've forgotten the dark days of Windows 95 or 98, where you would click on an icon, and have to wait 1-2 minutes for the OS to finish thrashing your HDD, due to not having enough RAM to run properly. (Or more recently: Vista or WIN7 on a 256 megabyte machine.)

Re:When do people get this

[dhavleak]

You mean, the fucking TFA aricle?

Re:When do people get this

[ultranova]

You'll excuse my ignorance, but from college I remember that usually you have 0-2V represent 0 and 3-5V represent 1. Does a 0 have a corresponding increase in amperage so that it levels out and uses the same amount of power?

If it did, then storing one zero bit would require 0-4 Watts and storing 1 requires 9-25 Watts. A full byte of ones would require 72-200 Watts, and a gigabyte around 77-215 gigawatts. Byt contrast, Hoover Dam has peak capacity of about 2 gigawatts.

In other words,I doubt it ;).

Re:When do people get this

[ArcherB]

Certainly under linux ram used as disk cache is marked "free".

Strange. On my system, it's marked as cache.

My System Monitor under gnome in Ubuntu 9.10 shows:
44% used by programs
11% used as cache

I'm not saying your system is wrong, just different. That's the beauty of Linux.

Re:When do people get this

[Barny]

To be fair, windows 7 will swap data out, my PC atm is sitting here with 8GB of ram, 1.1GB used by programs, 5.9GB used by cache and its reporting 1.2GB free, so it still pages out data. However I have never noticed it pageing, so likely its paging out the "right data", in other words stuff that is not used, just how an OS should work :)

Re:When do people get this

[John Hasler]

> Maybe the part about HDD caching slowing things down?

Yes. The blocks in cache are copies of ones on disk. When a process reads a file from disk the OS caches the blocks in otherwise-unused RAM. If they are still around when you need the file again they are right there in RAM so there is no need to go to disk. This speeds things up. If the RAM is needed for another purpose the blocks can be freed by simply changing a flag and mapping them as quickly as if they had been totally free. This does not slow things down. After a while most of the blocks you need will be in RAM when you need them. Thus caching makes the system faster, not slower.

The above describes Linux. I assume Windows works similarly.

Re:When do people get this

[Anonymous Coward]

FTFA:

Barth acknowledged that XPnet's data couldn't determine whether the memory usage was by the operating system itself, or an increased number of applications, but said that Devil Mountain would start working on finding which is the dominant factor in increased memory use.

In other words, it's all speculation on what's causing the increased memory utilization. I have to agree with the parent post that it's probably just Win7 finally making better use of available memory.

Why do I make this assumption? I'm assuming that the Microsoft developers, as much as I despise many of the technical decisions (or business driven overthrows of technical decisions), have at least a minimal level of competence. I am trying not to be arrogant and dismiss their decisions based on my 15 years in the industry working on similar code. I'm assuming that there are some bright engineers and developers working at Microsoft who are professional, not Microsoft drones, not Linux/Unix weenies (but understand the design of modern Unix/Linux), and are likely smarter and more handsome than me. I am giving them the benefit of the doubt.

Re:When do people get this

[jimicus]

Maybe the part about HDD caching slowing things down?

I could be wrong there, since I'm not an expert but I remember the dark, dark days when my computer when spend 2-3 minutes just to redraw a Word document. Why? Because it was using the HDD like memory, instead of using the actual memory. It seems to me that this problem, while minimized, has never completely gone away.

Anyway telling me "you're wrong" doesn't enlighten either me, or the other readers. Please elucidate.

There's a lot of variables, but in simple terms the theory goes that something which you have recently accessed (be it an application, a document or whatever) you are likely to want again in the near future. Hence it's worth keeping a copy in memory on the offchance.

On the other hand, you really don't want to be swapping. So if a program needs more physical memory than what you have immediately available, it makes more sense to allocate memory which was recently holding cached data and just reduce the cache size than it does to start swapping, which is what any sane OS will do.

If there's any real intelligence involved in this, the OS will re-allocate an area which hasn't been used in a while.

The cache would only cause a problem in the way you describe it if the OS did not dynamically resize cache to account for other demands on system RAM.

I can't explain the differences between yours and your brother's computer but I can tell you that OEM builds of Windows tend to have so much garbage loaded at boot that they often need serious work before they're genuinely usable. Some of the builds I've seen, it is a wonder they boot at all.

Re:When do people get this

[Anonymous Coward]

You were correct on one point: "the claims from the person who actually collected all the empirical evidence must be false"

Their tool is not capable of accounting for memory reserved by the system... it only sees the raw "available memory" which would be very little in a memory caching OS

Re:When do people get this

[Mr Thinly Sliced]

You are absolutely correct of course.

I've lost count the number of times the analysis done by application vendors / developers about the root cause of a performance problem under *NIX boxes is totally off base (not enough RAM for a process when they have stupidly mounted their log files under NFS etc) - as they aren't actually getting enough data to come to a correct conclusion.

Probably the case here. I'd be very surprised if Microsoft didn't spot such a glaring problem in Win7.

Re:When do people get this

[cgenman]

For the record, the researchers survive by making money through selling DMS Clarity Suite performance monitoring utilities to banks and other institutions. Which does not mean that they're wrong, but does reduce the potential objectivity. The CTO has as much invested in panicking potential customers about their system performances as an Antivirus company does about panicking people about hackers.

Re:When do people get this

[snemarch]

Wether the program is accurate or not is another matter, but the fact it doesn't report every system as using 100% of its memory suggests it is at least somewhat aware of superfetch etc.

Not really, SuperFetch doesn't necessarily use all available RAM - it depends on how much data it knows about... it doesn't just go out and read random files to memory :)

Re:When do people get this

[John Hasler]

> While yes, puppy linux may run super fast, its also lacking a lot of what
> makes people stay with windows ;)

Which is included in complete Linux distributions such as Ubuntu ;)

Re:When do people get this

[Mr Thinly Sliced]

You're right of course. I just lazily lumped non process mapped under "free" (since there is disk buffering too for inodes etc).

Re:When do people get this

[FlyingBishop]

I'm assuming you're joking, since I said nothing about amperage, and naturally if we're talking about an actual flip-flop unit in a stick of RAM, we're going to be talking about mA or even nA.

Though full Volts do sound a little excessive for small circuits, and in any case, it doesn't negate the basic point, which is that you have a smaller voltage for 0 and a larger voltage for 1 (or vice versa) and therefore you do have a low-power state if you're holding all 0's or all 1's (depending on which is low-power in your chip.)

Re:When do people get this

[kenp2002]

If only we could mod submissions... you are so right wanting a -1, Wrong mod.

First off there is nothing that correlates increased IO to the performance scenario.

IO != Bandwidth reqs.

First off if you want to increase your disk IO 10 fold, turn on security auditing and the last accessed time stamp tracking. Every time a file is scanned, indexed, etc additional IO is generated. XP vs Vista\Linux\W7 doesn't have any auditing turned on by default. No shit the IO goes up. There is a reason most Linux distros mount drives for desktops with a no access option to disable last access time stamps.

The backlog as they call it (Disk Queue Depth) is also now artifically higher as W7 actually behaves with NCQ\TCQ and a wide variety of disk enchancements. Buffering read\write instructions then reordering them artificially inflates the queue depth on a desktop as there isn't the volume of traffic to warrent a purge. That is why I turn NCQ\TCQ OFF on my desktops. It's akin to bitching about Raid-10 doesn't make maps in Counter-Strike load any faster. Only this dolt would re-invent the Raid is not for desktops\gaming machines argument.

For Virtual Memory since there is a higher number of standing services then XP I would expect more idle activity to be swapped out so % utilization of the swap file is meaningless unless you normalized that metric (e.g. boot the OS with no additional applications and record what % of the swap file is used normally when idle).

As any Linux user can state using as much of your RAM for buffers and cache is a good thing so long as you can grab it back for applications as needed.

32 vs 64 bit? 64 bit applications are inheritly larger. Compare two identically configured 32\64 bit kernels, the 64 bit is larger (I would assume twice as large but so far it only works out to about 15-20% larger so far that I've seen. I would guess due to byte alignment issues).

So far nothing in the article amounts to anything more then the ravings of a consultant that is more then willing to help for a small fee, get your "problem" under control.

Re:When do people get this

[GordonBX]

Which part is wrong?

Your old P4 is a minimum 3 - 4 times slower at raw CPU tasks than the Athlon X2 (http://techreport.com/articles.x/18448/13). The story about your 512Mb XP machine running faster than the Athlon X2 3Gb machine simply does not ring true. Perhaps you were only playing solitaire on it? No-one is denying that Puppy Linux is fast, or small, or that it fits in memory. Try running Open Office on it with only 30Mb of Ram and then tell me that it is fast.

Maybe the part about HDD caching slowing things down?

YES the part about HDD cacheing slowing things down. HDD cacheing speed things up! Memory Swapping slows things down - which is simply a question of how big your application + operating system's "working set" of memory is (i.e. the memory that simply has to be held in Ram because it is being actively read and written). 3Gb with Windows 7 is plenty for "normal" apps, and it won't be any slower than your 512Mb XP machine running the SAME applications - in fact it'll be a load faster because it won't have to swap as much and the disk is likely faster.

Re:When do people get this

[John Hasler]

> Does a 0 have a corresponding increase in amperage so that it levels out and
> uses the same amount of power?

It takes exactly the same amount of power to store a zero in RAM as it does a one.

Re:When do people get this

[Anonymous Coward]

:) May not be true for long. There are some recent technologies that allow not only processor but also memory quiescing. Doesn't apply to this thread, but I can see a time where unused memory doesn't use as much power.

Re:When do people get this

[Anonymous Coward]

Why would you install Vista or Windows 7 on a machine with that little RAM? Could you not read the system requirements?

Re:When do people get this

[hey!]

Well, sure. unused RAM is, in a sense "wasted". But it's not so simple as that. Sometimes having RAM in reserve in case *I* need it is a use.

Microsoft's *style* seems to be one of aggressive optimization. It's not a big deal to grab RAM, unless Windows has trouble giving up that RAM fast enough to serve *my* needs. That's why I stopped using Vista for development. I had 4GB of RAM and Vista would grab almost 100% of the 3GB available. Then I'd launch a process that needed 1GB of RAM and Vista would take an enormous performance hit.

What was worse it got worse over time. That was a clue. When I looked at the paging file, it had an unbelievable number of fragments, tens of thousands of fragments. Judging from the results, I'd say that when Vista realized it needed to give me so much memory it couldn't figure out how to release the RAM it was holding fast enough, and so it just started writing pages where ever the disk head happened to be. Dropping the page file altogether stopped the random system hiccups, but still didn't help the fact that Vista was crap at allocating large blocks of memory because it's not good enough at releasing the memory it grabs. So basically I just gave up on Windows and stuck with Linux for everything.

So you can't say *necessarily* that Windows grabbing unused RAM is innocuous. that depends on Windows being able to reallocate that RAM to *my* uses when I demand it.

The behavior I've observed is symptomatic of sloppy engineering. "This thing isn't fast enough, let's make every optimization we can think of." The first reflex of a good coder when confronted with code that is not fast enough is to make the code *better*, not *faster*. "Faster" often leads to unexpected costs, while "better" often leads to unexpected benefits.

Re:When do people get this

[Big_Mamma]

On any recent linux system, free reports only really free memory, not page/disk cache. That one is reported in the cached column. (sorry, no pre html tag @ /.)

free -m
total used free shared buffers cached
Mem: 48310 48090 220 0 94 15120
-/+ buffers/cache: 32875 15435
Swap: 16383 25 16358

PostgreSQL's performance depends on the page cache, so you can't see all of cached as free - if you let the cached number drop too much, your disks die.

Re:When do people get this

[hesaigo999ca]

ummm....can I see your credentials first before spewing out garbage?

Do you work for M$, are you a techie, do you even have more then one PC with multi configs
running multi apps concurrently, to make an experienced argument.

I can run windows XP on a 252mb P2 no problems, until i start to open too many apps, but till then it runs very smooth,
windows7 is the same architechture, just more bloated, with lots of wrappers for those previous wrappers, etc...
Why would it be any different. Did they change the files ystem all together, and use ext3, did they change the way they handle paging or caching....no.

Same rules apply here as in windowsXP.

Re:When do people get this

[snemarch]

Active RAM has the same periodic refresh cycle of idle RAM, but also the constant reading/writing of data over the system bus, which means about twice the energy used.

With your terminology, "active" meaning "being read/written"? You can easily have contents in RAM that's not being actively accessed, which means there won't be bus activity.

The question being: does simply having used RAM (i.e., set rather than reset bits) mean noticeably higher power consumption?

Re:When do people get this

[lagfest]

My current* windows 7 stats say:
Total: 2046MB
Used: 1.26GB
Cache: 634MB
Available: 743MB
Free: 132MB

So used RAM does not include cache. And 'available' is a nice way of telling grandma that RAM used for cache is actually available to apps.

* not a snapshot, i can't type that fast :)

Re:When do people get this

[John Hasler]

RAM that is marked as containing a block from a recently-loaded file uses no more power than would the same block marked as free.

Re:When do people get this

[snemarch]

Bull(2). The OS does not spend any time to "free up that RAM cache". the RAM is simply overwritten with new data, just like if it was free. The original data was just a copy of data from the HDD.

Almost correct - before handing out the memory to satisfy a memory request, Windows makes sure it's all zeroed out, for security concerns. Look at the various lists used by NT to organize memory pages, this link has a rudimentary description

Zeroing out pages is fast, though, and done with SSE write-through (i.e., no cache pollution). Iirc the SSE operations were done already in Win2k, and is definitely there in XP and later. So it's going to run at pretty much full RAM bandwidth... and outside of memory-starved situations, this zeroing happens in the background.

Re:When do people get this

[theaveng]

What if the programs don't "play nice" and refuse to release their RAM when it's needed? (i.e. Internet Explorer is storing a bunch of images you downloaded two days ago, and refuses to erase them from cache.) Wouldn't that force the OS to do HDD thrashing?

I understand that's not supposed to happen, but neither was Vista supposed to be a flop. I no longer trust MS to implement this stuff correctly.

Re:When do people get this

[Anonymous Coward]

Can we all please learn and understand the meanings of disk-caching and swap before we go down this alley any further?

kthxbye

Re:When do people get this

[asdf7890]

FYI, here's the memory usage of my Linux server: total used free shared buffers cached Mem: 720 702 17 0 55 510 -/+ buffers/cache: 136 583 Swap: 399 0 399 It says 702 MB of used memory. Now look at "-/+ buffers/cache", it says 136 MB. That's the amount of memory *actually* used by applications.

It isn't always that simple. Things other than just cache can get counted in the cached total too, the RAM allocated to VMWare VMs for instance, and and tmpfs mounted filesystems. For instance, running "sync; echo 3 > /proc/sys/vm/drop_caches; free -m" on one of my VM hosts shows 6479Mb reading cached still after the flush when in fact the vast majority of that isn't anything to do with data cached from I/O operations. This means that "free + buffers + cache" does not always give the amount of physical memory that could be allocated right now without needing to swap anything out - you need to flush the cache+buffers to see that.

Re:When do people get this

[snemarch]

actaully the windows 7 caching model is great. on games the difference between the first loading of a level and subsequent loads are night and day thanks to it's caching model.

That's the windows cache system generally, from way back in the NT days... Vista and later SuperFetch is more than that.

btw, regarding the article more directly: they shows no figure about the actual _swap_ usage, a thing that may or may not disprove their theory.

Indeed. The xpnet site does mention that they factor in paging somehow, but that's still pretty useless - paging activity needs to be a separate statistic. Also, simply looking at pagefile usage isn't terribly useful, an inactive app can have it's working set trimmed and pages flushed out to disk, and this won't matter much in the big picture.

What you need to look at is the rate of pagefile activity (ie., pages/second) as well as how often it happens - not just static numbers (even if having 1gig of data in the pf is probably a warning sign :))

Re:When do people get this

[lordlod]

You'll excuse my ignorance, but from college I remember that usually you have 0-2V represent 0 and 3-5V represent 1. Does a 0 have a corresponding increase in amperage so that it levels out and uses the same amount of power?

It seems you must have missed the complex electronics portion of your college.

5V TTL circuits use 0-0.8V Low and 2.2-5V High (on input), in between the high and low states is undefined. Regardless, modern RAM is almost certainly a 1.8V device externally and internally even less.

Modern RAM (DRAM) works by each bit of memory being a floating charge in a capacitor. When you read the bit the charge is released and read as either a one or zero. I wouldn't want to make assumptions about if a high voltage corresponded to a one or zero, they would choose whichever they felt worked best. This also includes whatever state they initialise the RAM to on power on.

In the ideal no-friction world, floating charge = no current = no power. In our world floating charges leak slowly and have to be topped up, so there is a degree of power being used depending on the high or low state of a given bit. That said, the refresh rate probably has more impact than the value.

In all honesty though, you are barking up the wrong tree. A far greater power issue in the modern computer is the increased power required by all the high speed external connections. Transmission line theory means that as the speeds of links like ethernet have increased the power required to shovel the bits down the line becomes exponentially linked to the speed of transmission. (EMF also becomes a serious issue: Fun game, wrap a GPS antenna in a ethernet cable then plug it in.) So to really save power you should start by unplugging your Gigabit link and hooking up some environmentally friendly 10BASE2 goodness.

Re:When do people get this

[value_added]

Apparently you've forgotten the dark days of Windows 95 or 98, where you would click on an icon, and have to wait 1-2 minutes for the OS to finish thrashing your HDD, due to not having enough RAM to run properly.

For yucks, install Win95 or Win98 on modern hardware and see how the GUI performs. If the near-instantaneous drawing of windows doesn't startle you, wait till you go back to using your spiffy new box for your regular OS and discover things work a lot slower than you thought.

Back to memory, though. I suspect one of the underlying problems with people understanding how memory is used is how Windows reports things. For most, I guess, that's a field or two in taskmgr.exe. For my money, the output of top on a Unix system is far more useful. From a FreeBSD system:

48 processes: 1 running, 47 sleeping
CPU: 0.0% user, 0.0% nice, 0.0% system, 0.0% interrupt, 100% idle
Mem: 63M Active, 714M Inact, 163M Wired, 18M Cache, 111M Buf, 39M Free
Swap: 990M Total, 124K Used, 990M Free

Doesn't require advanced knowledge to make sense of that ouput, or understand how memory is being used. And no third-party tools required.

Re:When do people get this

[jimthehorsegod]

That doesn't definitively state that they observed a rise in virtual memory use - it could as easily be taken to mean that they're extrapolated the memory usage and their belief is that this leads to virtual memory usage. For me, if they'd actually meant that it DID lead to virtual memory usage and that they'd measured such, I'd have expected to see numbers. In the absence, I'm inclined to believe that they see the memory usage and infer from that that virtual memory is the inevitable next step.

Re:When do people get this

[ByOhTek]

Ahh, but the person who made the comment didn't say Ubuntu, s/he said Puppy Linux.

Anyway, how much would it cost to build a system that could cache an entire Ubuntu install, with the mentioned features, to memory while still having memory to spare to run the programs?

Re:When do people get this

[Shakrai]

I think the issue here is that the system is turning to swap.

Turn swapping off. If you have a reasonable amount of RAM there's no reason to leave it turned on. I turn it off anyway for security considerations -- I use Truecrypt -- but really there's no reason not to do so on any XP machine with 2 gigs of RAM or Vista/Win 7 machine with 4 gigs of RAM.

Re:When do people get this

[Anonymous Coward]

until you meet a program which allocates tons of ram without actually using it - I hate to point out the jvm as culprit as I feed on java programming but that shit _never_ release back unused memory freed by the garbage collector.

Netbeans + JBoss + the stuff I'm actually building could easily eat 6gb+ of ram that could be swapped away as it's not being actively used, in order to make space for caching the file that netbeans is actually compiling at each step.

Yes, it sucks that the os has to come out with scheme for preventing stupid application ruining the system for every other application in there, but there is a limit to where a workstation ram could be upgraded - my company buy on cheap and cheap motherboards and 8gb is the max it could take.

Re:When do people get this

[Anonymous Coward]

the guy who is offering me numbers.

The guy admits he doesn't know what's going on, and doesn't know what is using the memory. If you cannot determine whether it's the kernel or running more applications, he's not speaking from an authoritive position, just as a confused user. That's a massive difference when it comes to making claims.

Re:When do people get this

[j00r0m4nc3r]

Using more RAM doesn't use more energy. Either your RAM is powered on, or it's not. And if it's powered on it maintains its contents, no matter whether the OS has actually written anything useful to it.

RAM has to be constantly refreshed or else it will lose state, so I can see how it could consume more energy to maintain a larger capacity. That's not to say that modern memory controllers don't just refresh all the RAM all the time, in which case you're always operating at the worst case, but I'd like to hope that they don't do that...

Re:When do people get this

[orange47]

hm, what about S3 standby. if it keeps refreshing DRAM for a long time, does it shorten memory life? (sorry for being offtopic, bu I need to know that). thanks.

Re:When do people get this

[Z00L00K]

Just because RAM is available doesn't mean that the OS should hog it. You may want to use that RAM for something different. It may be legal to use "excess" RAM for buffers, but then those buffers must be freed fast whenever necessary.

If you use large amounts of RAM for buffers you will either freeing the least used buffers and use them for the application, and then you will get memory fragmentation. This can be bad for some applications. The other scenario is that you will just kill a block of buffers and that may happen to be buffers that are heavily trafficked and has to be re-loaded somewhere else - and you have a performance penalty.

And caching a lot means more caching overhead. Not everything makes sense to cache. What about the case when you run a database, caching the database file through the OS first and then have the database engine to cache the same thing again? A complete waste of performance and resources.

So having the OS caching - it works fine for many applications, but not for every application. Some applications are also tuning themselves by looking at available memory and determining how to best allocate resources. It is of course possible to figure out how much the OS can free for application use, but it's also hard to calculate an usage budget. And if it goes wrong you will end up swapping to disk.

Re:When do people get this

[Z00L00K]

And just recently the hardware has been getting decent enough to run XP.

So when the hardware is decent enough to run Windows 7 we will need something else.

Re:When do people get this

[theaveng]

>>>You're making an apples-to-oranges comparison, and you don't mention what software is running on either of your machines...

Nothing exotic. The Windows OS plus Firefox browser. My P4-XP-512MB machine runs faster than my brother's AMD X2-WIN7-3GB machine. XP is more responsive.
.

>>>I've found Vista and Win7 to generally work better, giving decent enough hardware.

Well my Vista install was running Microsoft's minimum recommendation (512 megabytes) and ran like a snail through molasses. When I later upgraded it to 2 gig it did run better, but still nowhere as fast as my old 0.5 gig XP machine, which is why I eventually went back to XP.

My overall impression is that Vista is like a Beetle with a lot of chrome/pinstriping/et cetera. Looks good. Runs like crap.

Re:When do people get this

[drsmithy]

Certainly under linux ram used as disk cache is marked "free".

In top it appears as "buffers" or "cached". We have numerous servers with 32G of RAM, of which ~29G ends up being used as disk cache ("free" hovers around zero, and a little - maybe a few 10s to 100s of MB - of swap is used).

Re:When do people get this

[spxero]

...or the caching mechanism is broken.

I'm inclined to think it's this, at least for my Vista machine. I currently have 6GB RAM, but at any given time with Outlook, FireFox, and a handful of Explorer windows open there isn't any more than 2-3GB showing to be in use. The rest is cached. This becomes a problem only when I need to fire up a 2GB Linux VM for testing, the VM will pause itself on startup, citing not enough RAM available. I'm no expert, but I have a sneaking suspicion that the caching mechanism is the culprit.

Re:When do people get this

[slim]

What if the programs don't "play nice" and refuse to release their RAM when it's needed? (i.e. Internet Explorer is storing a bunch of images you downloaded two days ago, and refuses to erase them from cache.) Wouldn't that force the OS to do HDD thrashing?

Only if the program in question also insists on *reading* that chunk of memory.
Otherwise, it gets swapped to disk, then just stays there, never getting swapped back to RAM. No harm done, except that a small amount of HDD goes to waste until the next reboot.

Re:When do people get this

[FlyingBishop]

Thanks, that makes sense. Now I'm at a loss to explain why my machine runs hotter the longer I go without rebooting (bearing in mind that I hibernate regularly.) The obvious factor seemed to be used RAM, but maybe that's not it...

Re:When do people get this

[Anonymous Coward]

Definitely the part about HDD caching slowing things down. Even in the DOS age it was well known that hdd caching utilities (I forgot the names, too long ago)

It is smartdrv.

Re:When do people get this

[Anonymous Coward]

My config doesn't have any virtual memory configured, and it rarely reaches above 25% memory utilization, running Windows 7 x64 Ultimate.
Of course, I have 12GB of RAM installed.

Re:When do people get this

[swilver]

That's wrong. Turning off the page file can have very beneficial effects.

It is basically making a statement to the OS: "I donot want you to swap out Firefox/Word/Paintshop to disk EVER, even if I not touched them in two days as I may need them at a moment's notice".

The end result is that those programs remain snappy as they are always completely in RAM (as the OS has no choice in the matter).

If you donot turn off the page file, then the OS will begin wondering after a few hours of not touching those programs: "Should I perhaps swap these programs out to make more room for...."

  a) that blu-ray movie the user is watching (linear read) ?
  b) that download that is trickling in at low speed (but still runs fast enough to consume most of your cache overnight) ?
  c) random unpredictable file accesses (your machine acts as a file serer) ?
  d) that huge copy that is running in the background ?

Note that NONE of these actually benefit much from long-term disk caching... yet many operating systems will decide to try and make more space for the disk cache (by swapping out unused programs) if that happens to be the primary function of the machine for a few hours.

And then of course, when you go back to your favourite program... guess what, your machine is swapping again, despite having 8 GB of RAM.

Disclaimer: yes, I have 8 GB of RAM and swap is off for Windows because of the above issues. Same for the 4 GB Linux box.

Re:When do people get this

[maxwell demon]

What if the programs don't "play nice" and refuse to release their RAM when it's needed? (i.e. Internet Explorer is storing a bunch of images you downloaded two days ago, and refuses to erase them from cache.) Wouldn't that force the OS to do HDD thrashing?

I understand that's not supposed to happen, but neither was Vista supposed to be a flop. I no longer trust MS to implement this stuff correctly.

Is there actually an interface in Windows to tell programs "I need more memory, please try to free some"?

Re:When do people get this

[tepples]

You still have to load the same memory either way, and wait for the same delays.

Cached data does not have to be written back to disk, but running processes do. I think the complaint is that the virtual memory manager is incorrectly evicting processes instead of cached data. On some machines, such as some SSD-based laptops, writes are significantly more expensive than writes, so you want more reads and fewer writes. At least on Linux, an administrator can change the balance between the two by adjusting swappiness. Does Windows 7 have a corresponding tuning option?

Generally speaking you will get much better performance having RAM full of stuff you might need

Which falls down when the operating system incorrectly gauges what I "might need", thinking that I "might need" files that I copied two minutes ago more than the process whose window I'm about to bring to the front.

Re:When do people get this

[drinkypoo]

And just to swim in anecdotal waters, when I copied big piles of files to my GRiDPad 1910 via null modem cable using Microsoft's classic INTERLNK and INTERSVR for file sharing, SMARTDRV sped up the copy operations by about an order of magnitude. Life without disk caching isn't worth living.

Re:When do people get this

[Lumpy]

but really there's no reason not to do so on any XP machine with 2 gigs of RAM or Vista/Win 7 machine with 4 gigs of RAM.

Yes if you do the typical office no taxing tasks on a pc. If you do anything big it's no where near enough.

If you edit HD Video or very large Photo arrays you bump up against 4gig without effort. I hit the 16 gig mark on a regular basis.

Re:When do people get this

[Anonymous Coward]

Well there is that and the fact that most likely all of those people are still using a 32-bit OS which gives you a max of 3.2GB of RAM to begin with. As newer applications use more memory that 3.2GB of RAM isn't going to get you nearly as far either . Plus if you have a PCI graphics card the system is going to allocate memory to that as well further reducing that 3.2GB of RAM down to maybe 2.5GB of RAM.

Re:When do people get this

[snemarch]

YMMV - XP ran perfectly fine for me on an Athlon700/512meg ram... and flied when I upgraded to 1gig and disabled pagefile. Yes, Vista and Win7 require a higher baseline than XP, but in my experience they utilize powerful hardware better than XP. I obviously wouldn't install either on underpowered hardware, just like I wouldn't use XP (but rather Win2K) on even older hardware.

Re:When do people get this

[Lumpy]

$199.00 many of the acer net pc's have the specs to load an entire live Ubuntu CD into ram and execute the install while it's running.

2 gig of ram is dirt cheap and that is all that is needed.

Re:When do people get this

[tepples]

Anyway, how much would it cost to build a system that could cache an entire Ubuntu install

Ubuntu Desktop CD is 700 MB, and Ubuntu Netbook Remix is 1 GB. You can use this in your calculation.

Re:When do people get this

[theaveng]

Yes, I suppose you can run an entire system where it all goes into "RAM", and you'll see it as "more responsive" simply because you never have to touch the hard drive. But turning off HDD caching is a BAD idea. It makes cache misses that much more expensive because then, instead of having even the chance of finding what you needed in RAM or in the HD's onboard cache, you have to wait for the H/D to spin up and seek to the right sector.

I find this counter-intuitive. Let's say for example that you click on a folder icon, a window opens, and now all the "pictures" for the various icons must be loaded before the window can be displayed. This process will go a LOT faster (instant) if the pictures/icons are stored in RAM, rather than on a slow disk (several seconds).

You are saying it would be slower to load from RAM, and I don't see how that can be.

Re:When do people get this

[drinkypoo]

Compare two identically configured 32\64 bit kernels, the 64 bit is larger (I would assume twice as large but so far it only works out to about 15-20% larger so far that I've seen. I would guess due to byte alignment issues).

First, the kernel is compressed. Second, the compiler optimizes for your chosen architecture, and x86_64 is inherently more efficient. Third, most of the 32 kernels you're seeing are compiled for i386 or perhaps i586, and so they don't take advantage of most modern optimizations. Fourth, the memory use isn't greater except for data types which scale with architecture.

Re:When do people get this

[Anonymous Coward]

HDD caching does NOT slow down your system.
HDD swapping/paging does.

Re:When do people get this

[afidel]

Actually what you really need to do is calculate hard pages/second which for some retarded reason isn't available as a default stat counter, ie those pages which actually go to the secondary backing store (disk).

Re:When do people get this

[dzfoo]

I'd understand your confusion if English is not your first language. However, that sentence is explicit and unambiguous:

"resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks[...]" (emphasis mine).

The highlighted words assert that the slow-downs were a direct result of the memory consumption and that consequently virtual memory was consumed.

It is true that the article does not expressly state how they determined that virtual memory was consumed, but seeming that they make this assertion it stands to reason that it was at least observed in some manner. This is attested by a blog entry from the actual researches conducting the tests:

http://exo-blog.blogspot.com/

New data from the exo.repository shows that better than 8 in 10 Windows 7 systems monitored by the exo.performance.network are running alarmingly low on physical memory. And nearly the same number are demonstrating significant delays in I/O processing - ostensibly due to heavy virtual memory activity as Windows compensates for insufficient RAM. (emphasis mine)

        -dZ.

Re:When do people get this

[bdwlangm]

The disk cache is part of the OS, not the application. If an application loads some images from your HDD it's the OS that decides to cache them in memory, independent of anything your app does with them. Because disk caching is transparent to the app, the OS can free cache whenever it wants to.

Of course, historically IE would be a special case, as it was so tightly tied to the OS that the normal rules of transparent OS services didn't apply. Anyone know if this is still true?

Re:When do people get this

[salec]

Using more RAM doesn't use more energy. Either your RAM is powered on, or it's not. And if it's powered on it maintains its contents, no matter whether the OS has actually written anything useful to it.

That is not entirely true. Although much of the power dissipation in DRAMs comes from driving the data bus at high speeds (and it happens always, no matter what portion of internal memory space is actually used), some of it comes from memory matrix of the chip. Each DRAM cell is built around tiny on-chip capacitor. The capacitor can be either charged or discharged, which represents two values bit can assume. As bit held in the memory cell toggles its state, the capacitor gets charged and discharged. Whenever the capacitor is charged, it receives a portion of energy from power supply. When it is discharged, that energy is thrown away. The faster the average bit toggle rate and greater the average number of bits toggled, the greater the power dissipation of DRAM. Also, since those capacitors are "leaky" (they discharge spontaneously over time measured in milliseconds), a mechanism is in place to refresh them from time to time. Therefore, if memory contains static data consisting predominantly of bit states which correspond to charged capacitors, it wastes more energy then if it contains predominantly bit states which correspond to empty capacitors.

Re:When do people get this

[GooberToo]

I think the issue here is that the system is turning to swap.

Technically its page, not swap...but I frequently call it swap too.

Windows does things completely differently than most systems do. Windows uses the page file as a backing store for most everything. If you want to memory map a file, you actually map it into the page file and read it from there. Likewise, most things are automatically paged on Windows. That's why Windows requires a huge page file, and if allowed to dynamically grow, it will despite the fact it has enough memory for the job. Windows does this under the assumption that when it comes time to page, they can simply free the pages, as the associated pages have already been paged out. This often saves a complete write cycle. In doing so they have effectively alleviated the classic I/O bound problem which, for example, the Linux kernel authors have long struggled.

If you need to page, that likely means some portion of data also needs to be paged back in. In a classic implementation, you are now pushing data out to the page file, and as possible, freeing pages, which then allows you to locate the pages you desire, and page those into the newly freed pages. This has the result of causing high latencies and frequently thrashing the disk.

On Windows, this is not the case. On Windows, a more typical case is pages are freed, which then allows you to locate the pages you desire, and page those into the newly freed pages. This tends to create a read-only scenario rather than a stream of random reads, seeks, and writes which is the cause of the thrashing and high latencies.

While these are very different paging strategies, Windows also pays for it. It pays for it because for just about everything there is a paging cost, even if its a deferred write to the page file. In an environment where paging is not typically required, it means Windows is technically running slower than need be. Of course, this is also why many claim their system is noticeable faster when they disable paging on Windows. Conversely, this is one of the reasons why Linux and most Unixes are faster than Windows and one of the reasons why most all mainstream OS' designed to scale, scale far, far better than Windows.

When RAM was expensive and rare, I'd say Window's strategy made a lot of sense. Heck, I'd say it may of even been a superior solution, for the average user, compared to what Linux and most Unix's do. But now that RAM is cheap and plentiful, in my opinion, their "optimization" now only serves to slow systems down and confuse the hell out of people who don't understand how Windows uses paging.

The long of the short is, its normal for Windows to be paging - always - unless paging has been turned off.

Re:When do people get this

[theaveng]

>>>One point, turning off the page file is never a good idea, it means that the few things that can be paged out quite happily without negative effect stay in memory, taking up room the cache could use to make the rig run faster.
>>>

Then you just buy more RAM. - Or else accept that you won't be able to multitask 10 programs at once. You might only be limited to just 2-3 programs before RAM fills up. BUT you will have a more-responsive computer.

Re:When do people get this

[jimthehorsegod]

There is NO objective measure of 'slowing down' or 'significant delays in I/O processing' in the article at all, so I question why you're so certain that any such thing was unambiguously observed?

Re:When do people get this

[sopssa]

If the system is gratuitously using 95% of the RAM nearly all the time, then it’s a completely different scenario. Everything I try to open that wasn’t cached already will force the system to dump some memory to the swap file to make room for the new application.

Uh no. The point here is that the RAM is utilized with data that speeds up things, but that can be instantly freed if needed. It doesn't need to put that in swap file.

Re:When do people get this

[trboyden]

Then you'd be a fool who doesn't think profit is a motivation for a salesman lying about the usefulness of their product by blasting another's product to make theirs look good.

"...commercial DMS Clarity Suite..." indicates they have a commercial product they are advertising for, though I haven't been able to dig up any info on it or their clients that use it.

Re:When do people get this

[Rhys]

Turning to swap is only a problem if the system starts thrashing. My linux box with 4 gigs of memory, 4 gigs of swap is telling me I've got about 75 megs used in swap right now (600 meg file cache, 60 megs free main memory). It has probably had that stuff in swap for a week or more. If it isn't actually needed, no harm in it being off in swap. All the text console gettys I never see, for instance -- safe to swap out.

Re:When do people get this

[mdwh2]

Puppy Linux needs a whole 30MB of RAM? Talk about bloatware! I run AmigaOS on less than a single MB of RAM!

Re:When do people get this

[Anonymous Coward]

I think the issue here is that the system is turning to swap. Caching stuff that may be referenced again is fine and dandy, but if the system regularly turns to swap just to keep itself afloat, then you have a problem.

Windows, even XP, seems to use swap space almost immediately even if ram is available. I don't know if this is just a reporting error, but if you take a look at the task manager the page file is almost always being used, even if the available physical memory is more abundant than disk space.

Filling up a drive on Windows, even if there's almost no tasks running has some pretty scary results.

By comparison, my system monitor in linux almost always tells me that there's 0 bytes of swap used, and reports caching memory as free. It also appears to clean up after itself a lot better than if I close a task in windows...but that may just be appearances.

Re:When do people get this

[fast turtle]

If the idiots are looking at pagefile usage then I know they couldn't find their asses with a clue stick, map and a guide. On my system, my pagefile is constantly at 99 percent consumption because

1. it's a dedicated partition
2. it's set to the max size of the partition

According to Task Manager, I'm using just over 4GB of my 8GB of system ram (Win7-64) ad that's with Supreme Commander paused, uTorrent running, Word 2007, One Note and Outlook Open plus 30 tabs in Chrome.

ALL of this makes me wonder what their game is because none of it impacts me. I've yet to see my system hit the page file and I've been using WIn7 since the RC was released (now upgraded).

Re:When do people get this

[afidel]

Windows gets really cranky when it doesn't have a pagefile. We tried it for performance reasons and we saw an almost 40% drop in performance despite the server not being under any kind of memory pressure.

Re:When do people get this

[GuyverDH]

Why? Because it was using the HDD like memory, instead of using the actual memory.

That's not disk caching, that's virtual memory. Huge difference.

The only way to avoid using disk as virtual memory is, you guessed it (or maybe you didn't), have enough RAM installed to be able to disable Virtual Memory entirely.

That's where you went wrong.

You were whining about disk caching, when you were talking about virtual memory.

FYI - Virtual memory, aka swap in the Linux/UNIX world is something used by all modern OSes which might need more memory than is physically available.

Re:When do people get this

[Locklin]

The xpnet site does say they factor in "how often it relies on virtual memory", but not how they do this

Even that isn't ideal. It makes sense to swap out some library that hasn't been used in hours or days. The more ram available for disk cache the better. The only solution is to look at Memory used minus disk cache (like the 'free' command on Linux does).

Re:When do people get this

[slim]

Here's the confusion.

You think "HDD caching" means "moving stuff from RAM to HDD".

GP thinks "HDD caching" means "keeping a cache in RAM of stuff who's 'official' home is on the filesystem".

GP is using the more [ahem] conventional meaning.

You caused a lot of confusion by saying "HDD caching" when you meant "virtual memory" or "paging".

Re:When do people get this

[TheLink]

Yeah I don't know why they don't set up the counter by default.

Anyway to set it up yourself:

Start perfmon.msc
Then add counters
go to Memory, add "Pages Output/sec".

I'm not an authority on virtual memory but from what I know:
Page Faults/sec is not usually relevant for this - the virtual memory stuff will have page fault even if it's not swapping to/from disk - it's part of how virtual memory works.
Page Inputs/sec could happen when you launch programs (then the O/S starts paging in the stuff it needs) - it's no indication of running out of memory.
Page Output/sec on the other hand is when the O/S is low and needs to copy stuff in RAM and write it OUT to disk so that it can reuse that RAM for something else. This is the one you want to monitor.

Re:When do people get this

[maxwell demon]

Well, it's easy to get wrong - and considering Windows is used by regular Joes, it's OK that task manager simplifies matters a bit instead of showing up pages of detailed information.

There's nothing wrong with simplifying. However, given that the simplifying is exactly for those who don't want to learn the details, the simplification should give them the information which is most useful to them, and will best fit their mental image. In this case, I think a better solution would be to count cached pages as "free" memory (because after all, your programs can allocate them). An alternative could be to count unmodified cached pages as "free" (because they can be disposed with little extra cost), but modified cached pages as "used" (because they need to be written out to disk, just like when swapping out pages in use by a program).

Re:When do people get this

[idji]

Windows 7 Resource Monitor marks memory like this
"In Use" (2.97GB) (Memory used by processes, drivers, or the operating system)
"Modified" (122Mb) (Memory whose content must be written to disk before it can be used for another purpose)
"Standby" (944Mb) (Memory that contains cached data and code that is not actively in use)
"Free" (154Mb) (Memory that does not contain any valuable data, and will be used....)

screen shot

Re:When do people get this

[xigxag]

You missed the most important emphasis:

delays in I/O processing - ostensibly due to heavy virtual memory activity as Windows compensates for insufficient RAM. (emphasis mine)

Ostensibly means "to all outward appearances." In other words, they're admitting they don't really know the inner workings, the true cause of the delays. They're just supposing it's due to RAM swapping, as opposed to increased networking activity, aero glass, more concurrent programs being run on average, or any other number of other wag'd reasons. Basically, they picked two measurements that are both higher in W7, and just said, "Well, it stands to reason that A causes B." What's that phrase that internet smarty-pantses use all the time about this?

Re:When do people get this

[LinuxAndLube]

Although your post has been tagged as 'Informative', it is misleading. In Vista and Windows 7, memory used for caching shows up in the Task Manager as 'Cached', not as 'Free'. I must admit though that the adjective 'Cached' is a bit cryptic. How the hell are we supposed to know what it means?

Re:When do people get this

[growse]

I currently have 81 processes running on my Win7 install. Each one of them can address 2GB (I think) virtual memory. If you turn off the page file and a few processes decide to make full use of their addressable space, then everything will stop working. That's why a page file is a good idea.

Re:When do people get this

[TheKidWho]

Theres your problem, Firefox is a relic of a browser, use something like Chrome and you'll see the speed increase significantly. I've found Windows7 to be a speed demon personally.

Re:When do people get this

[Anonymous Coward]

I have the opposite experience. My Vista x64 system slaughters my XP system in performance, even though I have XP running on better hardware. On Vista when I run a program, it opens instantly. There is no hard disk grind and 10 second wait like there is in XP. Vista also doesn't have the terrible CPU hoarding XP UI where windows commonly "tear" or cause visual glitches while using up 100% processor time just for moving a window.

SuperFetch and the compositing window manager in Vista are what put it leagues ahead of XP's performance.

Re:When do people get this

[snemarch]

I kinda agree with you, but if MS did it that way, the *u*x zealots would run around in circles screaming lies & bloody murder. Catch-22? :)

Re:When do people get this

[Eivind]

Actually, even that is inaccurate.

You see, it can make sense for the OS to swap out some not-recently-used pages of a program, to free up more memory for caching. For example. Say you're playing a game, but you've got firefox open. It could make sense to page out the entirety of firefox, so as to have more physical ram free for caching of game-content.

Life ain't so simple in a virtual world :-)

Re:When do people get this

[Misagon]

Things can be even more complex and misleading.

The other day at work, I rewrote a program that does a lot of large disk accesses from using fread()/fwrite() calls to using memory-mapped files.
In both cases, the OS (Windows Vista) did the disk caching, and in both cases the performance was practically the same.

The difference was mostly in how the memory usage was reported. My program was reported as using 80% of all memory, instead of 5% that it was reported of using before.
Most of that memory is "cache" in practice, free for the OS to take and give to another program when it needs it.

Re:When do people get this

[Locklin]

jay@myhost:~$ free -m
             total       used       free     shared    buffers     cached
Mem:          2007       1900        107          0        208        722
-/+ buffers/cache:        968       1038
Swap:         4055          7       4048

Linux does swap some applications to disk, even when there is plenty of free ram. That is because it can be better to use the ram for cache then use it for an application that hasn't done anything for days.

Re:When do people get this

[Anonymous Coward]

Swapping isn't always a bad thing. Continually swapping things out and in certainly will hurt performance.

However, kicking stuff out of physical ram when it isn't being accessed leaves more for cache and other things.

Suppose I have a web browser open with an extremely animated gif heavy tab in it (takes tons of memory), I switch tabs and leave the browser open for several days doing other things. There's no reason that memory heavy tab should stay in physical memory. Eventually the OS will swap it out to disk. If I eventually do switch back to that tab, I may see tearing as it swaps back in, but in the meantime my system was a lot faster than it'd have been had that stayed in memory.

Re:When do people get this

[operagost]

The problem with articles like this is the lack of technical information is order to keep the PHBs engaged. If I knew that by "memory usage" they meant the total commit charge (see the Performance tab in Task Manager), I could assume the veracity of their results.

Re:When do people get this

[maxwell demon]

I'm assuming you're joking, since I said nothing about amperage, and naturally if we're talking about an actual flip-flop unit in a stick of RAM, we're going to be talking about mA or even nA.

DRAM doesn't contain flip-flops. It contains capacitors.
However I'm not sure how the number of ones and zeroes affects the power consumption of the refresh cycle. After all, an empty capacitor doesn't need to be refilled.

Re:When do people get this

[Antiocheian]

Then how does Vista-7 have a much slower hibernation compared to XP, running the same applications ?

If what you say were true, Vista-7 would ignore the cache before hibernation, and write only the essential data on the disk before shutting down.

Re:When do people get this

[snemarch]

Well, if you look at page activity per second, you'll get a good indication of whether the system is bogged down memory-wise. Trimming working sets might flush stuff to the pagefile, but won't give high continuous pages/sec rate.

Btw, "unused library" isn't best example - those can generally just be discarded and re-read from image, rather than having to be flushed to pagefile :)

Re:When do people get this

[fbjon]

Active RAM means memory that is actively used, meaning reads and writes, which take energy. It's still less than fetching from disk though.

Re:When do people get this

[del_diablo]

Firefox is bloat, and forcing idle stuff to disk cache results in chokedowns when the thing holding the screen stops.... :( Besides, a browser is a background application.

Re:When do people get this

[CrazedSanity]

The 'anecdote' to which you refer is completely plausible. Remember that the info wasn't based on any sort of raw processing power, frames per second, or disk throughput: it was based upon human perception. Don't ever throw the two together and think that they will always (or even mostly) match up.

I can tell you that a computer that has 4x the specs of another doesn't necessarily end up running better, regardless of the (Windows) OS. This can have a lot to do with hard drive fragmentation: one system that has low specs but is defragmented routinely can appear to run faster than a high-end machine that has 50% file fragmentation. I know it sounds like I'm picking on Windows, but the issue with data fragmentation--and the need to defragment--wholly resides on the shoulders of Windows users. And again, the speed difference, especially when it comes to perception, can be hugely affected by how much the hard drive is thrashing when a program is started.

Another issue in perception is with video performance. Two identical machines with the same model of high-end video card can have completely different apparent speeds if one has problematic drivers. A number of other things might be measured. I remember back when a person would say that their high-end, I-just-spent-more-on-my-computer-than-my-car computer was running slow simply because their dial-up connection was slow: the computer was running perfectly, but the user perceived it to be an issue with the computer (instead of the slow connection) because the data wasn't loading fast.

Re:When do people get this

[snemarch]

There's a couple of things to keep in mind wrt. memory-mapped files, though. First, there's the pagetable entries required for the mapping when you MapViewOfFile - these aren't "free cache". You'll be using roughly 4kb PTE data for each 4mb you map, excluding other OS overhead and 1st-level PT (I don't think mmf utilizes the extend 4mb or 2mb pages, but I could be wrong)

Second, you have to consider that you'll be introducing a lot of (cpu, not disk) pagefaults - not really a problem on modern machines, but still something to keep in mind; you'll get a #PF for each 4kb touched, whereas with raw file I/O you can get away with a ring3ring0 switch for each 64kb on XP-and-older, or 4mb (Vista) or even 64mb (2008 server) of data.

mmap sure as hell is convenient though, especially on 64bit :)

Re:When do people get this

[ByOhTek]

If you assume the base install is sufficient for most needs.

It's a little lacking in terms of office suits and web browsers.

Also, it contains compressed binaries (and other files?), which have to be decompressed at use, slowing down performance as well.

No, I don't think I can use that in my caclulations.

That being said, for basic use (web browsing, flash, office apps, etc.), I don't think 1-2GB is off the mark, so a system with 3GB of memory would be sufficient. 4GB of DDR2 (and not the cheapest available) cost me about $90 last week.

So, I guess I was being a bit knee-jerk there.

Re:When do people get this

[xlotlu]

Certainly under linux ram used as disk cache is marked "free".

No, it's marked as "cached". Try `free -m` in the command line: the "free" column reports the _real_ free memory. Many new Windows converts fall for that and go berserk about why their machine eats up all RAM, failing to read the 2nd line that presents the free memory including buffers/cache.

It was high time Windows started caching data from disk more aggressively, but if it starts swapping because of that, they obviously got it wrong.

Re:When do people get this

[mewyn]

With modern integrated circuits, especially those in a PC, voltage levels are typically about 0-1.85V, so you're off quite a bit there. Now, voltage with no current means no power consumed, think about a capacitor. If you charge it up and then disconnect the terminals, it will remain charged to that voltage, but there's no current (open circuit) so no power is consumed.

CMOS (based on MOSFETS) works similar to this. The gate of a MOSFET works on capacitance (technically electric field, hence the field effect in the name), and ideal CMOS circuits only use power when they are switched from one state to another (vs TTL which each gate will use power when in 0, but will not use power when in 1).

Memory is always active in standard PC systems. Sure, having more modules will draw more current. Memory modules typically will take a max of 1000mA when doing heavy operations, and at they typical 1.85V that most of these modules are, that's about two watts of burst power, but nominal draw is much lower. Overall two memory modules in a system is a small percentage of the overall TDP of a notebook system.

Re:When do people get this

[zippthorne]

I wish there was a tool that could tell you how many cycles are wasted waiting for stuff that's not in each of those caches, so you'd know what to try to upgrade next. No point in throwing more RAM in if the bottleneck is insufficient L3 cache...

However it was my experience with windows up to XP that turning off swap does unfortunately result in a performance improvement. My suspicion was that it was preferring to put things in there that are actually in use in favor of caching stuff from the disk that more rarely gets used, and possibly doing it at inopportune times.

Re:When do people get this

[Eskarel]

Yes, but the only evidence we have that anything you "do need" is being swapped out is presumptions from someone who rights crappy overpriced performance monitoring software.

They don't say that Windows 7 generates ____ page faults, they just say memory was highly utilized which is not necessarily a bad thing.

I've never seen this behavior in Windows 7, nor have I ever heard of anyone who has. If it was really causing swapping on 86% of PCs we'd have seen something about this before now. It's just as likely that the monitoring software has a memory leak on Windows 7 as anything else.

As a side note, hasn't the SSD myth been debugged by now? Last I checked the math, writing the volume of the disk every single day would give you thirty years before the drive died and it scaled in a linear fashion. Even noatime probably isn't actually necessary on an SSD.

Re:When do people get this

[Anonymous Coward]

haha, very funny!

Re:When do people get this

[amliebsch]

You don't need third party tools in Windows, either. The Performance Monitor admin tool will let you track virtually any metric you would want to.

Re:When do people get this

[tepples]

Also, [the live CD] contains compressed binaries (and other files?), which have to be decompressed at use

Is decompression slower than seeking and reading?

Re:When do people get this

[theaveng]

"Firefox is a relic of a browser, use something like Chrome"

.

Uh huh. You remind me of my redneck pal who says anybody who drives anything other than Dodge is a fool. He bases his opinion on..... well basically nothing, and I bet you do too.

IMHO brand loyalty is silly. Look at the people who loyally stuck with Toyota..... and where it got them now. (Failing old-sludged engines, sticking accelerators that drive the cars off cliffs, and now priuses with prematurely-dying batteries.) Today's great company will eventually be tomorrow's hasbeen, and it looks like Toyota has fallen off that cliff. As GM did, Toyota sacrificed quality to churn-out more cars and become "#1".

BTW tests show Opera 10.5 is the fastest browser, not Chrome.

Re:When do people get this

[Bakkster]

But more page-faults doesn't always correlate to more slowdowns. An OS with better page-allocation prediction will run faster (from the user's perspective) with the same number of page-faults. It's only a problem if the page-faults are on cached data that the user is requesting at that moment.

Continuing the Firefox example: it might be one page of memory to each page you want to view. A smart OS will leave the pages with the main Firefox program and current tab in RAM and cache the others first. Then when tasks switch, the cached Firefox pages are reloaded while the user is still looking at the first page. There are page faults, but the user experiences fewer delays.

Basically, there is no meaningful conclusion we can draw just from peak RAM utilization and page fault numbers (either average or peak). To do that, we would need to measure the number of page faults for pages that were already in memory but cached to disk and that required the user to wait before continuing.

More importantly, to claim that Windows 7 is 'bloat' just because it uses more RAM and has more page faults is erroneous without additional evidence.

Re:When do people get this

[jedidiah]

My Ubuntu installs average around 5G.

Re:When do people get this

[jedidiah]

I remember those days. That's also when memory prices dropped enough such that
you could get a whole 32M or more. Once Windows was liberated from the HDD being
a memory bottleneck it ran remarkably faster.

It was a stark enough difference that some people might mistake it for a CPU upgrade.

Re:When do people get this

[Anonymous Coward]

Uhhh something about correlation and causation me thinks.

Re:When do people get this

[tepples]

They don't say that Windows 7 generates ____ page faults

I've seen programs generate "page faults" in Windows Task Manager even though the hard disk light is off. I think it has something to do with having done Windows's equivalent of calling mmap() on /dev/zero.

As a side note, hasn't the SSD myth been debugged by now?

I didn't mean "expensive" as in wear but as in time. The 4 GB SSD that came in my Asus Eee PC was horribly slow at writing, so I replaced it with a 32 GB RunCore SSD. I apologize for my ambiguous comment.

Re:When do people get this

[Gr8Apes]

I've played with the actual backend WMI bits on 2008 R2 (same core as W7) and can tell you there is no such thing as an easy way to make these numbers appear. The "Used" number you see is actually the total - available. The actual numbers reported for the apps added together frequently exceeds the actual total RAM, no matter which available stats you use.

I'm not sure how accurate the "available" number is either. The free number appears to be the only one that's truly accurate and verifiable.

Re:When do people get this

[TheLink]

If stuff slows down due to that swap out, then it's still accurate enough for me.

Maybe the O/S could get it right and swap out and swap in Firefox in a way so I won't notice any slow downs. Give me an example of such an O/S please.

So far in my experience, if Windows or Linux swaps out Firefox for whatever reason, if I then switch to Firefox, I have to wait for it to be swapped back in.

Why "page out" and not "page in"?

"Page in" doesn't necessarily mean that I'll have to wait if I switch to different programs- the O/S is bringing stuff from disk to ram - I believe in some cases the O/S pages in stuff as part of running a new program - so it's not such a useful metric for "not enough memory".

But "page out" means something in RAM is going to disk - if I ever want it back in RAM, I'll have to wait.

If stuff in RAM is going to disk needlessly and causing unnecessary waits then the O/S virtual memory algorithm is getting things wrong.

Re:When do people get this

[VGPowerlord]

That's the windows cache system generally, from way back in the NT days... Vista and later SuperFetch is more than that.

It must have changed at some point, because Windows 7 64-bit is considerably snappier at loading programs the second time than the first. Vista 64-bit had nowhere near the speed doing that as Windows 7 64-bit does.

For example, I play Team Fortress 2 a lot. The first time I load it (with Steam already running, so that load time isn't a factor), it takes about 45 seconds. If I close it, then open it again later without rebooting, the load time is approximately 5 seconds.

Now, I have 8GB of RAM in this system, so I have lots of RAM I'm not using at any given point, so it can keep programs like this in memory.

P.S. These times are just the time before the initial switch to Full Screen mode, there is a slight pause for loading after that as well, but that pause isn't any shorter or longer than before.

Re:When do people get this

[Anonymous Coward]

I agree. I have 8 gigs of ram and I never use all of it in Windows 7. However, turning off the page file does give me a noticeable performance drop.

Re:When do people get this

[thePowerOfGrayskull]

Here's what I've noticed - and it's not just in Win7, it also happens to a lesser extent in XP. I do think the memory is being used for cache -- but Windows will cache things to the exclusion of memory actually in use by applications. So if you have minimized something to the taskbar, or just haven't used it in a while, you will almost *always* have to wait while it gets loaded back from the page file. In the case of a large application (IDE, 3d editor, etc) this can take many seconds. On a quad-core machine with 4GB of RAM, I just don't see that this is acceptable.

I think it's great that Windows is using all available RAM for cache - but it should not do so at the expense of running applications because that RAM is *not* actually available.

Re:When do people get this

[theaveng]

>>>then everything will stop working. That's why a page file is a good idea

They won't stop working. Windows or Linux will give you the option to turn-off some of those processes, so you can only run ~40 instead of 80.

BTW don't you 81 is a little excessive? I've only got 4 running right now (because I want the machine to run fast, now slow).

Re:When do people get this

[Foofoobar]

Incorrect. Using 95% of any given resource is a BAD idea because it does not take into consideration usage spikes. For instance, should we run our web servers with 95% RAM usage? Should we run our databases servers at 95% RAM usage? So why is a desktop any different? It can get spikes in usage as well from the user/s.

True the spikes will be fewer as a desktop but the system is there for the user and not vice versa.

Re:When do people get this

[thePowerOfGrayskull]

I don't know how caching works in W7 but on Linux, if the system has to "turn to disk-based virtual memory to handle tasks" then the memory utilization isn't caused by buffers because buffers are never swapped out to disk. If W7 behaves in a similar manner the it's either a memory leak, system bloat or the caching mechanism is broken.

Not entirely true either. I've seen swap usage grow (albeit very little) when there is over 2GB in use only by cache.

Re:When do people get this

[Xest]

To expand on what the parent is saying, basically Windows sets a block of memory for each process aside and changes this on peak memory usage. This is a somewhat simplified explanation, but if an application uses 20mb of RAM 99% of the time, but then spikes to 200mb of RAM that other 1% of it's running time, rather than resize the pool each time from 20mb to 200mb, it'll just leave it at 200mb after it's noticed this pattern a few times to avoid having to keep enlarging, and shrinking (and as the parent said- zeroing out) the additional memory required.

If Windows gets a bit short on RAM, it will not bother with this optimisation, and will just do the resizing each time, to ensure that real RAM is available as much as possible- because the resizing is still faster than paging with virtual memory.

Effectively, it's just a free performance boost for when there is enough memory available. Sure it looks bad in task manager, but it's not really.

As I say though, the above was a simplified explanation, the real rules on when memory does and doesn't get allocated and the pool for each process does and doesn't get resized I believe are based on a rather more complex algorithm than that, so it is potentially possible that part of Windows 7's speed optimisation involve modifying this very algorithm, hence why things may look different in task manager now than they used to.

I've not personally noticed any issues with Windows 7 bar one, with Firefox, where I went to bed one night and left it running, coming down the next morning to see it eating up a massive 1.8gb of RAM. I suspect this really was a memory leak and am inclined to blame Firefox seeing as it seems to have been on a downward spiral into shittiness ever since about version 3.0- I find it tends to crash sometimes now also which it never used to, all this since I updated to version 3.6. I didn't find previous versions had these sorts of issues, but I have found the browser getting progressively slower with each version. I even contemplated switching back to IE, until I tried browsing with it and got a reminder that whilst yes, Firefox seems to be getting worse, at least it's still not _that_ bad ;) Chrome was an option, nice and fast but the UI just pisses me off.

Anyway, I digress. Their assertion of slow downs based on what appears to be merely theory from them extrapolated from a misunderstanding of Windows RAM allocations seems to fly completely in the face of the general global consensus that Windows 7 actually performs very well, particularly compared to Vista. So I suspect their entire conclusion- the theory about virtual memory usage and slow downs is all just the result of their initial flawed premise of how Windows allocates memory.

One final note, I was looking for an article on MSDN I read a few years ago that gave a better description of that I have described above with a slightly more in depth explanation than I've been able to give off hand, but I can't for the life of me find it now. if you are interested in more detail, it may be worth having a dig around for it to see if you can find it though as it was quite a good article.

Re:When do people get this

[Anonymous Coward]

Certainly under linux ram used as disk cache is marked "free".

It most certainly is not. Actual snapshot from one of my servers:

Mem: 4147184k total, 3976700k used, 170484k free, 47804k buffers
Swap: 6291432k total, 328k used, 6291104k free, 2780212k cached

The disk cache is found as "2780212k cached" in the above, and is very much not marked as "free" at all. That would be the "170484k free" figure and is a rather volatile thing due to processes starting and ending, and thus freeing up the memory they allocated for themselves.

Please note that under Linux (and most modern operating systems) with any actual load on them, one of the goals of the operating system in typical configurations is to use as much memory as possible to cache data read from disk in order to not have to read it again, since unnecessary IO kills overall performance. Various algorithms of wildly different complexities are applied in order to decide what to cache, in order to do as good a job as possible.

Also note the very low figure for allocated swap (328k in my example) which means that the machine is happily chugging along at its production load (hovering at ~2.15 or so at the moment) with approximately 65% of its memory used to cache stuff and no active swapping going on.

It's a very healthy machine, using all of its memory, as it should.

Re:When do people get this

[sopssa]

It doesn't matter as the cache can be quickly cleared from the RAM to make room for whatever app needs it. Every modern OS does this, not just Windows.

Re:When do people get this

[Foofoobar]

First I said ANY GIVEN RESOURCE... not just RAM. Second, disk cache is not as fast as RAM and should not be relied upon like RAM. It is merely a safety net if you run out of RAM... like when you get a usage spike.

Re:When do people get this

[hellothatsme]

For people who are disturbed by this: There is an easy way to fix the problem. I took out one of my two memory modules and now my system uses never more than 50% of my RAM.

Re:When do people get this

[ShadowRangerRIT]

If you have a reasonable amount of RAM there's no reason to leave it turned on.

If the swapping algorithm is so bad that it swaps unnecessarily, then yes, turning off swap will help. But a good swapping algorithm remains useful even if you have 16 GB of RAM. Large sections of many processes are basically "run once, then ignore" or even "never run". Most processes have a decent amount of startup code that is never referenced after the first half second of execution, or load multi-megabyte shared libraries into process memory space to get two short functions (or, similarly, contain code that only runs under exceptional circumstances). If you disable swap, you're denying access to that memory not only to other programs (which we'll assume for the sake of argument don't run out of RAM themselves, since you use the phrase "reasonable amount of RAM"), but for I/O caching.

If you've got a program regularly crawling part of your directory structure, or you're writing frequently to files, the RAM freed by swapping out the junk parts of each program could be used for a productive purpose. Delaying the write means you can write a contiguous block all at once, allowing higher priority I/O to go through without forcing the low priority I/O program to block, and it can also mean reduced fragmentation on the disk.

Similarly, a predictive caching algorithm can improve responsiveness with that otherwise wasted RAM you've decided *must* be kept in memory. If you always start a program around 6 PM when you get home, the system can recognize this from the metrics and preload it. If you don't run it, oh well, the RAM was used just as effectively as if it held unused, unswappable contents. If you do use it, your program starts nigh instantaneously. If the program itself has a specific performance profile, where specific data files are predictably read, the computer can cache them using the RAM freed by swap, reducing loading times and increasing responsiveness. That might seem wasteful (spinning up the hard disk when it isn't needed), but in other cases it saves energy; if you're seeding a torrent, and the computer has enough memory, it may cache the whole file in memory; voila, no matter what piece a client asks for, the disk doesn't need to spin up.

That said, no swap algorithm is perfect. And if you've got 4+ GB of RAM and all you're doing is running a browser, an office suite and maybe a game, the difference will be small (or non-existent if all your programs and all reasonable file system caching can fit in memory). This doesn't mean swap is useless; all it means is that it's not perfect.

Re:When do people get this

[Mister Whirly]

You don't have any idea if it is XP that is faster, or some other hardware/software component. If you really want to test it properly, you need to have 2 hardware platforms that are identical - not an Intel and an AMD system - have the same software on each, and run actual benchmarks that give you hard data, not just "mine seems faster than my brother's". Anything else is just opinions and anecdotal trivia.

Re:When do people get this

[Anonymous Coward]

And that is exactly why I marked this article as a dupe. Also, congrats for your comment, +1 Insightful.

Re:When do people get this

[nedlohs]

They understand it perfectly. They just want you to install their shitty application.

Re:When do people get this

[TheKidWho]

I never said Chrome was the fastest. But compared to Firefox it's a speed demon. Firefox seems to be single threaded while Chrome uses multi-threading for the tabs. For example I can play multiple flash videos in chrome or browse and watch a video with no problem. In Firefox, if I two windows open, one playing a flash video and I start scrolling on a webpage on the other video, the video starts to lag and skip frames. Not to mention websites load significantly faster and the browsing experience is much better. Firefox was good for it's time, but it's a relic now that needs some major under the hood improvements.

Mind you, this is on a 4GHz Core i7 920 system with 6gb of low latency DDR3-1600 RAM...

Uh huh. You remind me of my redneck pal who says anybody who drives anything other than Dodge is a fool. He bases his opinion on..... well basically nothing, and I bet you do too.

Thanks for calling me a redneck and accusing me of making baseless accusations. Nice try, however I'm a white Eastern European and an Engineer.

Re:When do people get this

[ShadowRangerRIT]

You might try establishing a small, fixed size page file. At least in older versions of the OS (not sure about Vista and later), the crankiness over the lack of a page file could be solved by creating a simple fixed size 2 MB page file. Of course, given that there are large parts of the OS that are loaded into RAM, run once and never touched again, making it slightly larger would free up some active RAM for other processes and cache, so a 50-100 MB page file would not be unreasonable.

Re:When do people get this

[Anonymous Coward]

In my linux cache is marked as cached. Ubuntu 9.10

Tasks: 176 total, 2 running, 174 sleeping, 0 stopped, 0 zombie
Cpu(s): 20.9%us, 0.6%sy, 43.4%ni, 34.9%id, 0.2%wa, 0.0%hi, 0.1%si, 0.0%st
Mem: 8194064k total, 8143728k used, 50336k free, 96000k buffers
Swap: 24009100k total, 11972k used, 23997128k free, 6737524k cached

Re:When do people get this

[Bobfrankly1]

Mod parent up. I going to post the same if I didn't find it in the comments. Glad to see I'm not the only person who reads.

"This is alarming," Barth said of Windows 7 machines' resource consumption. "For the OS to be pushing the hardware limits this quickly is amazing. Windows 7 is not the lean, mean version of Vista that you may think it is."

Barth acknowledged that XPnet's data couldn't determine whether the memory usage was by the operating system itself, or an increased number of applications, but said that Devil Mountain would start working on finding which is the dominant factor in increased memory use.

Barth contradicts himself within 3 adjacent paragraphs. He has no idea what is causing the increased memory usage. What is alarming is how often people will believe sensationalistic stories like this one.
Of course, I'm sure companies such as Corsair and Kingston don't mind.

Re:When do people get this

[ShadowRangerRIT]

One other minor note: Windows's use of pre-emptive paging makes for a much faster hybrid sleep and/or hibernation. If your page file is larger than main memory, and you're not paging excessively, most of your memory is probably already paged out. Thus, the hibernate file only needs to have the unique data written to it; on a laptop with 4 GB of mostly used RAM and a relatively slow hard disk, it could take two minutes to hibernate the machine (hope your battery lasts). Every bit of memory paged out preemptively means less time to hibernate. My home machine is set up for hybrid sleep, and has 4 GB of RAM. Time from issuing the sleep command to hibernation is about 3-5 seconds, and that only works because of the page file.

Re:When do people get this

[VGPowerlord]

BTW don't you 81 is a little excessive? I've only got 4 running right now (because I want the machine to run fast, now slow).

er... he said processes, not applications. Windows XP SP3's Task Manager shows 5 applications and 48 processes running on my computer at this very moment. Plus the application count is wrong, because it doesn't count ones that don't currently have a Window (like Outlook).

Heck, Windows itself uses more processes than 4.

Re:When do people get this

[ShadowRangerRIT]

For laptops, Windows' strategy still makes some sense. As I noted above, if memory is pre-paged, then a hibernation can occur nigh instantaneously. If it isn't, you've got to wait for X / Y seconds, where X is the amount of RAM in use, and Y is the write speed of your hard drive (typically slow in a laptop). If you've got 4 GB of mostly used RAM, and the hard disk only writes at 40 MB/s, that means hibernating takes over a minute and half. It raises the risk of head crashes substantially (since most people just close the lid and go, and every jostle during this write is another chance for disaster), and can interfere with waking it up again if you change your mind.

Re:When do people get this

[Shakrai]

If you edit HD Video or very large Photo arrays you bump up against 4gig without effort. I hit the 16 gig mark on a regular basis.

Then buy more RAM. Somehow I don't think you'll be editing HD video very successfully from the swap file.....

Re:When do people get this

[jwinster]

I'd like to see some metrics comparing hit rates from XP to Win7... Just out of curiosity to see if their algorithms for disc caching have improved.

Re:When do people get this

[Anonymous Coward]

You forgot to tweak your config.sys. Send them all to Himem dude!

Re:When do people get this

[Anonymous Coward]

Nice post. Just a little comment: L1/2 cache is not only neccesary because of the clock multiplier. The RAM being at the same clock speed than the CPU does not mean that you can fetch data from RAM in one clock tick. In fact, there where 486DX and even 386 computers that included L2 cache in the motherboard, which was much faster than RAM.

Re:When do people get this

[snemarch]

Things [i]have[/i] changed since XP :) - by default it's cache system is pretty conservative. You can get around that by setting LargeSystemCache=1 (except if you're using ATI drivers - nastier-than-normal BSODs galore!). Then 32bit and the kernel/user address space split becomes the limiting factor.

With more physical RAM, a larger address-space, the cache has more room to play - and it doesn't hurt that Win7 (like Vista) is less conservative than XP was :)

Re:When do people get this

[kenp2002]

Your ram is refreshed on a regular basis to maintain the information (All that CAS and other timing metrics). Regardless of the electrical requirements, lets just use 1 unit for 0 and 2 units for 1 over the course of a few seconds the normal distribution of 0s and 1s is going to pretty much give a consistent draw of X amount of electricity, used or otherwise.

Modern OS systems tend to randomize where in memory applications reside now and with buffer overflow boundaries and randomized data in unallocated space you can pretty much just pick a given draw of power to the stick of ram regardless of the actual usage. Once there is an OS running you can pretty much disregard per-bit power consumption with conventional ram. Perhaps with solid state and phase change type memory architectures and perhaps the memristor may change that dynamic such that it is something the OS will need to manage but from a security standpoint I'd rather have random crap in unallocated space rather then all 0s. It makes identifying unused memory harder to spot to a rogue application and from a termal standpoint I'd rather have consistent temperatures across a stick of ram rather then inconsistent temperatures because we have no data being stored at the time.

Re:When do people get this

[Omestes]

I'm even more leery. I'm not entirely sure that the type of usage that their basing this on is common, or generalizable to other scenarios.

Anecdotally, I currently have 3 computers running Win7 in my household (and know of several others) and none of them have this problem with normal use. My main computer (Win7 Pro, AMD PhenomII x4 3.4ghz, 6GB RAM) runs around 20% utilization in normal operating conditions, and peaks around 60% for heavy use. I have managed to get above 90% but very rarely, and only while doing tasks that are intrinsically heavy on memory (having 12 images open in photoshop, playing with browsers in ways not intended).

This is probably not a common problem, but only relates to one usage scenario. Once again Slashdot's editors take a specific problem and try to blow it into a huge cloud of FUD.

The headline should be "In Application x, Windows 7 maxes out memory, according to person y's usage."

Re:When do people get this

[Lumpy]

Actually yes I can. I am typically editing the clips that are in ram. I hate it when I load a clip to cut down and it fails to load because I need an extra 1.2 gig of ram and being a weenie I disabled swap for no real reason.

P.S. call me when you can buy a laptop that has 16 gig of ram. YES I do edit in the field at times in a paltry 4 gig.

Re:When do people get this

[Man Eating Duck]

Why would you prevent your operating system for accounting for the possibility of filling all that RAM?

To enable it to use that ram for file cache. I tried to disable the pagefile when I upgraded to 4GB a couple of years ago, and I experienced a distinctive increase in loading times for games. This guy knows more than I do and explains it better:

If you've got plenty of RAM in your PC, and your workload really isn't that huge, you may never run into application crashing errors with the pagefile disabled, but you're also taking away from memory that Windows could be using for read and write caching for your actual documents and other files.

There are other reasons as well, the article is well worth reading.
BTW I'm on Linux when not gaming, and have a similar experience from that OS as well.

Re:When do people get this

[shutdown -p now]

I can't find the link now, but I recall that at some point Linus also said something to the effect that you are better off (perf-wise) with a swap file even if you seemingly have plenty of RAM to fit all your working processes at any given time. This was about Linux, of course.

Makes me wonder why that is the case. Not that I don't trust Linus on such things (especially when other OSes behave similarly), but it's extremely counter-intuitive.

Re:When do people get this

[recoiledsnake]

Even if assuming that more used RAM takes more power, any savings would be absolutely nullified and exceeded by the power taken to spin up the hard disk for information that the RAM would have cached.

Re:When do people get this

[prefect42]

Certainly under linux ram used as disk cache is marked "free".

No it isn't.

I'm here sat on a machine with hardly anything running, and I've got less than 150Mbytes of 'free' memory. Ignoring anything else, I've got 2254596kB Cached and 365836kB buffers. That's disk cache/buffer, and it's not listed as Free.

jh

Re:When do people get this

[Big Boss]

FWIW, I run Firefox on most of my machines, but on my Netbook (1Ghz Atom) I run Chrome. FF would peg my CPU at 100% just sitting at a /. page, or other heavy Javascript/AJAX type site, even Gmail. Chrome hovers at about 5% doing the same stuff. I love FF, particularly with Weave as I use multiple computers, but they really need to get their crap together for smaller spec machines. That Chrome, Safari, and Opera can do it tells me that FF is the problem.

And yes, I tried removing ALL plugins from FF, even while leaving some installed in Chrome.

Re:When do people get this

[Daengbo]

You mean your machine with 8GB RAM never hits swap? Wow. Shock. Color me surprised! O_o

Re:When do people get this

[jellomizer]

no no you got it backwards. If Linux does this it is because of good memory management. If Windows does it, it is because of bloat.

Re:When do people get this

[Anonymous Coward]

Turn swapping off. If you have a reasonable amount of RAM there's no reason to leave it turned on. I turn it off anyway for security considerations -- I use Truecrypt -- but really there's no reason not to do so on any XP machine with 2 gigs of RAM or Vista/Win 7 machine with 4 gigs of RAM.

So I guess you're the Windows eqvuivalent of a Gentoo user??

Re:When do people get this

[operagost]

Windows uses a memory model that is similar to OpenVMS. There is a reason it is called a pagefile, not a swapfile. Eliminating the pagefile has a different, negative impact on this kind of system.

Re:When do people get this

[Anonymous Coward]

Sounds more like explaining virtual memory / swap space to me.
When the CPU accesses a valid memory address, these things happen:
Is it in L1 ?
Is it in L2 ?
Is it in L3 ?
Is it in physical RAM ?
Invoke OS routine that frees up some physical memory and loads it from swap space / pagefile on disk. This will usally non-blocking, so the OS switches to some other task, until the loading of that page is done. The request to load that page is also probably not fullfilled immediately, but queued somewhere, since that drive is already busy with some other stuff. So that might all need a lot time.

If physical memory is too low to store all the memory pages that are regularily needed by some tasks, you'll end up very often accessing stuff that's not in physical memory, but in pagefile, so you end up with heavy disk activity. The os needs to drop some important page from memory to load another one from disk, so that task has to wait, shortly afterwards one of the pages that was dropped formerly is needed, so needs to drop another important page to have physical mem to load the one that's needed, task has to wait, and it goes over and over and all you hear is rattling of your harddrive and not much happening.

Re:When do people get this

[Moryath]

The answer is: Windows (up to XP) had an incredibly crappily coded structure for handling HDD caching (which ought to result in improved performance) and Swap (which only improves performance in certain instances).

One of the old workarounds, believe it or not, was to format your system in a very specific way: create a very small (~5GB or so) partition right next to the cylinder dedicated to nothing but the swap file, then a second partition "outside" onto which the OS went. The idea being to keep the Swap file (accessed very often) in a confined and predictable space, rather than letting it spread itself in fragments all over the platter. Some people even took it to the logical extreme of buying a small-capacity but high performance drive (say, a 10-20GB 7200RPM UATA133) to keep their swapfile on while using a larger but slower (and less expensive), say 5400rpm 130GB platter, for the OS and their files.

Re:When do people get this

[Anonymous Coward]

This reminds me of an old employer who was constantly ragging clients to run their machines at 100% capacity.

Until one day she said it in front of an IBM engineer who pointed out that if you're routinely running at 100% then you have absolutely zero reserve capacity for when the system comes under stress. Which on the system in question virtually guaranteed a hard fail of the entire OS.

Re:When do people get this

[Orange Crush]

I'm not sure if it matters anymore with more recent versions of Windows, but from force of habit all the way back in the Win3.x days, I've always set the minimum and maximum swap size numbers to the same so that the pagefile remains a fixed size on the disk to keep it unfragmented. Does this still actually help anything with Windows XP/Vista and 7?

Re:When do people get this

[TheLink]

Whether Linux or Windows, I still keep swap around on most machines.

The idea was for my machines to not run out of memory immediately but get slightly slower and run out of memory before things get extremely slow.

So the swap size I pick is something like "maximum number of seconds I'm willing to wait" * typical disk speed when swapping.

However if you have a laptop running Ubuntu, it seems the swap file is used for hibernating as well so if you want the hibernate feature you need a swap file that's large enough to fit the RAM. On Windows the swap file is separate from the hibernate file, so you don't have that problem but you do have a different problem if you are low on storage space and actually want some swap (like say a netbook with an 8GB SSD ;)).

In the end, I may still have to turn off swap on Windows XP since it often complains about low virtual memory even if I have lots of free RAM.

Re:When do people get this

[Just Some Guy]

You caused a lot of confusion by saying "HDD caching" when you meant "virtual memory" or "paging".

Virtual memory and paging aren't the same thing. VM maps a location in a process's address space to a physical memory location. Two processes can both think they're writing to the same address, but instead be writing to a totally different location. This makes it basically impossible for a program to trash another's storage (yay security! yay stability!) and makes long-term memory leaks impossible (delete that memory map and the resources are automagically returned to the system).

Paging is a nifty hack that you can easily implement once you have a good VM system. If a process tries to access a memory address that isn't physically loaded into RAM, you pause it, copy the appropriate page from swap to RAM, update the process's memory map to say that the memory location refers to that new page you just copied up, then resume the process.

You can have each without the other. For instance, you could implement paging without virtual memory if you don't mind having all your processes share the same flat memory space. There were Amiga products that did exactly this. And even if you don't care about paging, virtual memory is enormously useful, to the point that it's not much of an exaggeration to say that modern computing would be impossible without it.

Re:When do people get this

[jthill]

IBM mainframes at least used to have "expanded storage", block-addressed and fast enough that pagefaults backed there were handled synchronously. It made a good canary metric, "pagefaults across the main memory boundary" though it wasn't reported with that description.

Re:When do people get this

[DarkOx]

Editing video should work fine out of page file. Even the oldest of paging prediction algorithms should do a good job with keeping the working set in memory for such a sequential task. Unless your are doing your editing on a system you have badly mis-configured in your own ignorance.

All resent versions of Windows 2000 or later should do a great job of memory management for video editing with the default automatic settings; irrespective of if you have 4 gigs of memory or 40.

Re:When do people get this

[Foolhardy]

But "page out" means something in RAM is going to disk - if I ever want it back in RAM, I'll have to wait.

On Windows it doesn't necessarily mean that. Writing a page to disk != needing to read it back from disk later.

Each process has a working set. Pages in the working set are mapped actively into the process's VM with page tables. The memory manager aggressively trims these pages from the working set and puts them into standby memory. A page in standby is not mapped for reading (and more importantly for writing) anywhere in the system. Part of putting the page into standby involves writing a copy to disk. This will show up as a page written.

From standby, the page can be used one of two ways:

1. Transitioned back. If one of the processes that originally had the page mapped touches the page, it will cause a soft page fault in which the page is simply put back in the process's page directory. There's no need to retrieve it from disk since it still has the same data from before. The disk copy is discarded. This will show up as a transition fault in the performance monitor.
2. Reused for something else. Standby pages are counted as "Available" because they can be immediately re-used for another purpose without accessing the disk. The memory copy of the page is discarded and the page is re-used for something else. No disk activity is needed at this time since there is already a copy on disk. When one of the original owners of the page want the data back and the page is no longer on standby, it has to be retrieved from disk. This will count as a page fault in the performance monitor.

The nice thing about this model is that disk activity isn't needed to either reuse pages or bring them back at the time of the demand. It helps avoid the ugly condition of paging one process out while paging another in at the same time, causing disk thrashing.

Since Vista, the memory manager will preemptively re load pages that have been bumped out of standby back into standby if there is free unused memory available. Also since Vista, each page of memory has a priority from 0-7 that determines which pages are preferred to keep in RAM. In all versions of NT based Windows, memory mapping is very similar to page file management and will use many of the same counters (including standby memory, transition and hard faults, pages in/out). Memory mapping is used by lots of components internally and for loading executable images and libraries. Also, file caching is logically based in many ways on memory mapping, although the counters are different in many cases.

Re:When do people get this

[nabsltd]

If you want to memory map a file, you actually map it into the page file and read it from there.

Unless this has changed drastically in Vista or 7, this isn't exactly how it works.

When you memory map a file, it does do so by making it part of the paging system, but it doesn't copy the contents to the system paging file. In essence, the Windows kernel merely adds the memory mapped file to the list of current paging files on the system,..it's just private to anything that has the file memory mapped.

Re:When do people get this

[Anonymous Coward]

It's worth noting that it isn't OEM builds of windows, but OEM installs of windows (and restore disks) that include a bunch of shovelware/crapware. Also, running slow and poor performing AntiVirus tends to affect performance more than anything these days. Norton 2009+ or ESET's NOD32 work well (I'm not a big fan of NAV, but the recent versions are about the best performing, they did a ground up rewrite around 2008).

Re:When do people get this

[ztransform]

Windows gets really cranky when it doesn't have a pagefile. We tried it for performance reasons and we saw an almost 40% drop in performance despite the server not being under any kind of memory pressure.

And yet when I turn off swap on my 32-bit Vista laptop performance increases 1,000 - 10,000% easily. The difference between waiting 10 minutes for the computer to stop thrashing the swap file and near-instantaneous action is immeasurable.

Typical "don't turn off your pagefile" responses are fraught with lack of experience. There are times when a system performs better with a pagefile. There are many times when a system performs so much better without a pagefile that one wouldn't dream of ever turning it back on.

Re:When do people get this

[ztransform]

SWAP is using the HDD as extra RAM when the system doesn't have any more memory left to use

Windows also copies memory to the swap file (pagefile) during periods of inactivity so that it can quickly flush the main memory for new requests.

Re:When do people get this

[Blakey Rat]

The three classifications below the "memory used" indicator in Task Manager (in Vista) are:

1) Total (obviously, your total amount of RAM)
2) Cached (the amount being used for the disk cache/DLL cache)
3) Free (the amount completely 100% free)

Right now, my "Free" is 9 MB, meaning that my computer is using 99.78% of its 4 GB of physical RAM. However, of that, 2109 MB are used for various types of Cache and can be free instantly for application use, if an application requests it.

The problem here obviously is that this company is counting the "Cached" amount with the "Used" amount, in which case it'll almost always read 100% or close to it. That's how the system is *supposed* to work... to say that indicates a problem is just retarded.

It wouldn't surprise me that win7 has a heavier memory footprint though - as more applications move to .net and web browsers use lots of flash / silverlight etc - all of these things have a RAM cost.

This is Vista, which (according to Slashdot lore at least) is more bloated than Windows. I have 1.77 GB being used, and here's what the task bar shows is running:

The OS itself/Explorer
Outlook
Windows Live Messenger
Firefox (about 15 tabs)
SQL Management Studio (this guy's a beast)
IE7 (about 5 tabs)

That's all in 1.77 GB. If I go ahead and open Visual Studio, it pops up to only 1.86 GB. (Probably because VS shares a large portion of its code with SSMS.)

In any case, considering what my system is doing, I find this amount of memory use perfectly acceptable. And that's on the "bloated" Vista.

Re:When do people get this

[MSTCrow5429]

Turning swapping off is a horrible idea, and will likely lead to an unstable system. Swapping cannot be completely avoided, irregardless of the amount of actual RAM present.

Re:When do people get this

[cbeaudry]

Wow, you really do believe everything the media says.

Toyota is in NO way different than GM, Ford, Chrysler, VW, etc...

Only difference is, the USA is on a smear campaign against Toyota and all import vehicles.

Why? Toyota is the biggest fish in the pond, the the other big fish want some room to grow.
They are knocking Toyota down a peg.

Whats happening with Toyota and the recalls they are having are no different then the dozens of recalls GM, Chrysler and Ford EACH have every single year.

Recalls are about numbers. They always have been, they always will be. And Detroit has been an expert at that game for the past 50 years.

The media just doesnt want to remember that.

Re:When do people get this

[Blakey Rat]

Probably not. NTFS gives special priority to the pagefile, it gives it a really wide berth, so it's unlikely to get fragmented no matter what you do. (Unless you have a runaway process that goes *really* nutsoid, possibly.)

Re:When do people get this

[X.25]

RAM is wasted when it isn't in use. The fact that the task manager in Windows says your RAM is used 95% tells nothing, and no it won't "result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks". I'm actually really surprised, and not in a good way, that "chief technology officer" of the company doesn't know this.

Of course that RAM is wasted when it's not in use.

That's why Linux does it properly, and uses RAM.

Unlike Windows 7, which uses RAM and then continues using your Virtual memory (read: hard disk), because it's buggy.

I hope you understand the difference.

Re:When do people get this

[bonch]

Ladies and gentleman, I present to you the "texting" style of writing.

Re:When do people get this

[Jackie_Chan_Fan]

yup. Good old smartdrv. It helped a lot.

Re:When do people get this

[Anonymous Coward]

"
If theres a sudden need for more RAM, the cached data can be "dropped" in no time. It doesn't matter if it averages at 25% or 95%, just that the perfomance overally is better when you utilize all the resources you can to speed up things in general.
"

Remarkably, this rarely if ever happens. Since Linux 2.4+ and Windows 2000+, it has been impossible to restrict the amount of "system cache" or "cache" that is utilized. Prior to that, yes.

It -was- possible in 2.2 and some 2.4 kernels. It -was- possible in Windows NT. You could arbitrarily restrict the amount of system cache the memory manager would use, out of total system memory. The feature was removed since 2.4, as apparently having more options is bad (/sarcasm).

A design goal of a so-called "good" memory manager is to use all available system memory. It's not a good goal under all conditions, but it is A goal. Apparently everyone who's ever written a modern memory manager cannot conceive of the concept of allowing the user to tune their system.

The problem
-----------

Once all memory is allocated to cache, it takes a significant (as in, user perceptible) amount of time to release said memory for application use. You can test and demonstrate it right now on Ubuntu, Fedora, Windows XP, or Windows 7. The methods to create this massive performance nightmare have already been enumerated in this thread, so there's no need to rehash that.

Back in the good old days of 1992, it was possible to drastically increase the performance of either DOS or Windows with a nifty little application called Smartdrive. This reserved a portion of memory and dedicated it for disk cache. Not a lot, just a little, but it increased performance in some cases by 10 times. This memory COULD NOT be used by Windows for anything. It was dedicated to one task.

Fast forward to 2010. Now it's dynamic! ooo aaa! Now it's efficient! ooo aaa! Now it creates more delays, frustration, and negative performance for the typical desktop user than any other aspect of modern operating systems. (IMHO) boo hiss! " Why is my system slow? " Oh, right, because cache = available memory. Riiiiight.

Change creates the illusion of progress.
In this case, this concept of "we must use all RAM, zomg, we MUST!" is a step backwards. Until this ridiculous notion is reconsidered, all desktop users of modern operating systems will suffer the consequences. (note, desktops, not servers)

The solution:
-------------

Allow a tunable parameter in both Ubuntu and Fedora, and a registry setting in Windows 7 that permits a portion of system memory to be reserved for caching operations.

OR

Allow the user to RESERVE a portion of system memory that will NEVER be used for caching operations. This will actually then be available when applications need it.. You know, when you start Firefox and stuff like that.

Then you'll see systems scream. Until then, enjoy your mediocre performance.

Re:When do people get this

[Blakey Rat]

Doesn't require advanced knowledge to make sense of that ouput,

I'd argue that knowledge of what "wired" means in that context is advanced knowledge, required to make sense of that output. I've been working with computers for a hell of a long time, and I have absolutely no clue what "wired" means in regards to allocation of memory.

I also don't have whatever knowledge (advanced or otherwise) to understand the difference between "buffer" and "cache." And I'm only guessing that "inact" means "inactive", although I don't really get the distinction between "inactive" and "free."

My first guess would be "wired" is total amount of RAM, but from the numbers that's clearly wrong. In fact, that report doesn't seem to tell you your total amount of RAM at all... which is less than useful.

Anyway, the point is, if you think that readout is obvious, you've completely lost contact with the general public. I hope to God you don't write UIs that you expect other people to use.

Re:When do people get this

[Anonymous Coward]

I don't have any "learned something today" crayons.

Re:When do people get this

[julesh]

Wether the program is accurate or not is another matter, but the fact it doesn't report every system as using 100% of its memory suggests it is at least somewhat aware of superfetch etc.

I believe Windows tries to maintain a pool of free memory that isn't being used by the cache. For instance, my system (which has been up for nearly 3 days straight now, and used heavily in this time, so there's more than enough time for the cache to have filled available memory) currently has 750M commit, 660M cache, and 60M kernel, with 1570M total available, which means 60M is sitting there unused.

Re:When do people get this

[kenp2002]

you miss read, all that aside, due to word and dword alignments, even under compression an executable on 64 bit architecture may have to send a 64 bit wide instruction rather then 32 bit wide so executables often will simply be larger due to padding issues (tech and arch differences aside.) I was suprised by the fact they weren't defacto twice the size but the amount of code that must be fixed width (and byte aligned) is less then I expected (I would assume compiler optimizations.)

Re:When do people get this

[IndigoDarkwolf]

Wait, wait, wait. Did I miss something? Since when was defending Windows the hip and trendy thing to do on Slashdot? I thought Microsoft was the root of all evil, was utterly incapable of anything approaching competence, and that the sooner Linux replaced it as the consummate desktop solution across the world, the better?

Re:When do people get this

[growse]

Let me guess: System, System Idle Process, svchost and winlogon.exe?

Re:When do people get this

[julesh]

And just to swim in anecdotal waters, when I copied big piles of files to my GRiDPad 1910 via null modem cable using Microsoft's classic INTERLNK and INTERSVR for file sharing, SMARTDRV sped up the copy operations by about an order of magnitude. Life without disk caching isn't worth living.

A few years back, I had to install XP on a machine whose BIOS refused to boot from CD. The only route I could work with was install DOS, then run the XP setup program from DOS. But I forgot to install smartdrv; when I came back the next morning it was still in the initial "copying files to the hard disk" phase (i.e. before the first reboot). Stop it and install smartdrv, the process finished in less than a minute. Don't know why it was so slow, probably writing the files 1 byte at a time or something...

Re:When do people get this

[Reziac]

Fixed size helps but better is to give it its own partition.

Myself, I just turn the damn thing off, and have done so since Win98. Only reason I've ever found to have a pagefile in a high-RAM machine is that some Photoshop plugins whine "not enough memory" if they don't find one.

Re:When do people get this

[Anonymous Coward]

I presume the last sentence is tongue in cheek but I think the Slashdot favorite for that one is "Correlation is not causation" or some variant thereof.

Re:When do people get this

[Anonymous Coward]

Bullshit. Opera 10.50 only has the fastest Javascript performance. That says nothing about HTML/CSS rendering, image rendering, SVG or a whole load of other factors. I use Opera 10.50 beta as my primary browser and the development build of Chromium as a test platform and "quickie" browser, so I speak from experience. I would like Opera to one day outperform in all aspects, but if I'm honest then the overall speed crown still goes to Chromium.

Re:When do people get this

[Anonymous Coward]

He's surfing porn while he's writing. Difficult to use CAPS one-handed.

Re:When do people get this

[Anonymous Coward]

ram doesnt show 95% its always a number out of a number. the 95% you are reffering to is the processor usage and yes its very important that stays low.

Re:When do people get this

[GargamelSpaceman]

There are things that also use an ssd keyfob for disk cache and others that use one for swap so as to speed this sort of thing up. b

Re:When do people get this

[Anonymous Coward]

Don't know where you got your information from... but page faults are bad news any day and regardless of heuristics (because let's be honest, they're not perfect and never will be) less RAM available always means more page faults. Fetch time for the RAM usually involves several thousand CPU cycles. Fetch time for the hard drive usually involves several million. Do the math - RAM is slow, retrieving from the hard disk is even slower! Just because we have the RAM available doesn't mean we should use it! Developers should always strive to build software with as small a memory footprint as possible without compromising performance. Unfortunately, Microsoft has never really understood this simple fact.

Re:When do people get this

[QuantumRiff]

You are right.. But how much power does it use if you have to spin the HD and seek to your data compared to using electricity for ram cache

Re:When do people get this

[hairyfeet]

Uhhhh...you act like that is a big deal for some reason, why? I paid a grand total of $60 after MIR for 8Gb of RAM (four 2Gb sticks of DDR2 800MHz), so it isn't like that is crazy expensive anymore. RAM is cheap, what is crazy is not maxing out your board when it gives such an incredible performance boost.

As for Windows 7, been running HP X64 since release day and I have to say it is one of the fastest OSes I have ever run. When I first turn it on while I am surfing it will continue to access the HDD for around a minute and a half as it caches all my favorite programs (I have noticed it even seems to learn the difference between what I use during the day and what I use at night) and from that point on it is lightning fast.

Now the gadget I use shows actually RAM usage and NOT the cache, and so far the highest actual RAM usage I have been able to hit is 1200Mb, but if I go to perfmon or launch process explorer it will show that about 80-85% of my RAM is being used for cache, which is REALLY fine with me. I absolutely HATED how WinXP would let huge chunks of free RAM just sit wasted while it would head for swap. I paid for the RAM, it is still drawing power, use it you &%^$$&$ OS! With Windows 7 I can launch all my favorite apps nearly instantly, and when I launch a large one I don't use everyday like World In Conflict, then it'll dump the cache and give it to the game. Smart, really smart. And it has learned I game at night and surf daily, as the games run faster at their usual times played than if I decide I have a free minute and launch in the middle of the day.

So I have to call bullshit on TFA. Most likely they are simply using perfmon to look at free RAM, which doesn't make any sense anymore. Why let all that RAM just sit idle when it can make the PC so much more responsive? A much better metric would be something like hard page outs, to see how many times it is having to go to the HDD. I can say once my PC is up and running I don't see the HDD indicator turn on unless I am saving something. Much better IMHO to actually use that big pile of cheap RAM instead of letting it sit. And RAM is cheap, if he would have said say 32Gb than I would have agreed with you. But today even a cheapskate like me can afford 8Gb fairly easily. And if the board will hold it, why not get it?

Re:When do people get this

[Albert Sandberg]

Hehe, rather have a "+1 wrong" so I can learn from that experience :-)

Re:When do people get this

[FlyingBishop]

No, if you look around, I'm off base - that's what happens when you get a BA and don't learn anything about modern hardware, only the hardware from the 80's.

But I'm thinking about the case of a computer (like my desktop) which has 4GB of RAM, and if I really need that much RAM I'm likely doing something that's going to flood the whole memory space and cause massive paging anyway - like if I'm loading up the art for a game or something. Once the game is closed, there's no reason to hold all that shit in memory, since even if I play the game again, there's no way of knowing if I'm going to need the same textures, etc.

Also I was thinking about mobile stuff, where all the memory is flash anyway, so you have a ready supply of memory that definitely costs less than RAM.

On the other hand, using flash as RAM isn't the best of ideas thanks to its lifespan.

Re:When do people get this

[Anpheus]

If you use truecrypt, use full disk encryption or set the policy option to zero out the page file on shutdown, or numerous other things you could do.

Without a pagefile though, there will be some random quirks that you'll have when using Windows. I don't know how similar it'd be to running linux without a swap partition.

Re:When do people get this

[Danse]

Continuing the Firefox example: it might be one page of memory to each page you want to view. A smart OS will leave the pages with the main Firefox program and current tab in RAM and cache the others first. Then when tasks switch, the cached Firefox pages are reloaded while the user is still looking at the first page. There are page faults, but the user experiences fewer delays.

Ok, I'm truly curious here. How the heck would the OS know what memory space corresponds to the active tab in FF versus the inactive tabs or other application data?

Re:When do people get this

[scot4875]

Disclaimer: I also have 8GB of RAM, with 64 bit Win7. I also have swap enabled.

At any given time, I generally have Opera, Firefox, and IE all open, usually with at least a dozen tabs open (one Hulu tab is always open for the Daily Show). I also have Visual Studio open, my IM client, a virtual machine or two, and at least a few other applications open. I run WoW with all of this stuff in the background. I have a dual monitor setup, and frequently do stuff on the secondary monitor with WoW open on the main one.

I never have any of the issues you seem to have. I see no swap partition utilization, and no performance issues when switching between applications. And no, this isn't a case of me not knowing how slow my system really is because I don't know what to look for; I know what a slow system feels like and what a fast one feels like.

Yes, you are correct that none of the operations you list benefit from disk caching. I copied a 10GB VM image between drives last night, and it had zero impact on the rest of my system's performance.

Disabling swap *might* gain you a tiny bit of performance in some cases. In most cases, it really won't. And in many cases, having no ability to swap will be a big limitation on what your system can do for such a small benefit.

--Jeremy

Re:When do people get this

[Danse]

Since Vista, the memory manager will preemptively re load pages that have been bumped out of standby back into standby if there is free unused memory available. Also since Vista, each page of memory has a priority from 0-7 that determines which pages are preferred to keep in RAM. In all versions of NT based Windows, memory mapping is very similar to page file management and will use many of the same counters (including standby memory, transition and hard faults, pages in/out). Memory mapping is used by lots of components internally and for loading executable images and libraries. Also, file caching is logically based in many ways on memory mapping, although the counters are different in many cases.

What controls that priority setting? I'm asking because it seems like it might relate to a question I asked in response to another post here. It seemed to me that the only way that the OS would be able to intelligently cache the memory used by FF in order to keep the necessary app data and current tab data available would be with some direction from the application itself. Some sort of prioritization scheme would be needed for that. Doesn't sound like that's the sort of priority that you're referring to here, but I was wondering if there might be something I'm missing.

Re:When do people get this

[srvivn21]

Certainly under linux ram used as disk cache is marked "free".

Depends on where you look. The output of "top" (and indeed /proc/meminfo) shows only unused memory as free; the output of "free" has a separate line that shows used and free memory "-/+ buffers/cache".

Re:When do people get this

[swilver]

Trust me, that rarely happens. With 8 GB, you'd need to have 4 of those processes. Chances are that they will be killed (for using too much memory) before that.

Re:When do people get this

[BlackSnake112]

What video card(s) are installed? If the win 7 machine is using crappy video (which is likely with a store bought system) The XP machine will feel snappier. Does the win 7 machine have on board video or a real add on card? I say real since I have seen a few of these PCIe DVI cards that are just a converter from VGA to a DVI card. The video processor is still the on board chip using system RAM.

Prove it, run a few benchmarks. Run a few of the gamer and business ones then report back. Those numbers might surprise you. Or they may surprise us. No way to tell until they are run.

Also, the vista recommended RAM was 1GB not 512MB. 512MB was min needed to run. Also in those recommended spec was a 128MB dedicated video card. I have added a good ($70-$100 cheaper now) video cards to a number of systems. Nothing else was done. The end user said the computer was faster. Also windows 7 != vista. Win 7 runs better on a computer that ran vista ok. It has on the 45 machines that I have installed it on at work anyway. I get the feeling your comparing XP to a 2006 vista system with built in graphics. When vista came out none of the on board/built in graphics chip sets could run it snappy. That is one of the major vista sucks it is slow arguments. And they were right. The built in graphics really did not have what vista needed to run correctly.

Re:When do people get this

[Bobfrankly1]

No offense but I'm going to side with the guy who appears to make his living testing these sorts of things ... the guy who is offering me numbers.

Please remember that 95% of statistics are made up, and in this case, the guy offering numbers doesn't really know what 1% of them mean. He also doesn't mention how many actual windows 7 or windows xp machines he has in his testbed.
He just throws out 23,000 plus machines. His website states they monitor workstations and servers. How many are windows 2000, vista, or a version of the Server OS? He could have 20,000 XP machines and be comparing them to 10-100 Windows 7 machines.

I'm all for useful data, but I don't think this guy is offering any. I declare shenanigans.

Re:When do people get this

[scot4875]

I'm just going to put this bluntly: you do not understand how memory works in a modern OS.

An application has no responsibility for things that happen with the swap/page file and disk cache. That's entirely up to the OS. Applications gain their benefits/drawbacks by simply running on the OS.

--Jeremy

Re:When do people get this

[swilver]

If no swap is being used, than you obviously won't have the problems I mentioned.

For me however I noticed a definite improvement when it comes to leaving my system running overnight when swap is off. Everything is instantly up and responsive still in the morning (or next evening, or next week).

It's possible though Win7 was improved in this regard. Perhaps it no longer will slowly swap programs out under disk cache pressure.

Also, and this is something most swap-proponents donot seem to understand: using more memory for long term disk cache (as opposed to prefetching) is only beneficial if it is caching data that:

  1) is actually accessed more than once (caching a big disk copy is therefore usually pointless)
  2) is not being requested at speeds slower than harddisk speed (ie, internet traffic, CPU bound processes that read files, movie playback, music playback, etc).

It is also pointless to increase a 6 GB disk cache to 7 GB by swapping out half your programs. If the data didn't fit in 6 GB, it is likely it won't fit in 7 GB either.

So, having the system make more room for the disk cache by swapping out programs has to be for a VERY good reason. Maybe they did get something right in these newfangled Windows versions :)

Re:When do people get this

[Bakkster]

Ok, I'm truly curious here. How the heck would the OS know what memory space corresponds to the active tab in FF versus the inactive tabs or other application data?

Most recently accessed would be a good way to determine recent tabs, and most frequently accessed (or the one where the program counter is pointing) gives you the main executable. Just start by caching the least recently and frequently accessed pages first, your necessary application data to reload should be the last thing out of RAM. QED.

Re:When do people get this

[Ed Peepers]

Not anymore. On my newer Win 7 system (multicore, 8GB RAM) I can happily play Mass Effect 2 with uTorrent, Virtual PC, Outlook, and Firefox active in the background and Alt+Tab around when I feel like it. I could never pull something like that off before. I have no illusions that it's all thanks to Win 7 -- it's the shiny hardware letting me do this -- but Win 7 is in no way impeding.

Re:When do people get this

[AmiMoJo]

A 32 bit OS can only have 4GB of total virtual memory, which includes both physical RAM and page file. In fact it also includes address space used for PCI(E) devices and the like.

Re:When do people get this

[AmiMoJo]

You can't actually disable paging completely anyway. Even if there is no page file the OS can still page things like programs and DLL files to disk, because the .exe or .dll contains a complete copy of the data. Of course there is no actual writing to disk, but the physical RAM does get freed and the data loaded in again later.

Re:When do people get this

[BitZtream]

There are plenty of reasons to leave swap on. Unless you're running every application you have from ramdisk already, swap can provide a performance boost by paging inactive tasks and freeing the ram for caching.

There are MANY reasons for having swap, turning it off is an almost certain way to slow your system down unless you know exactly what you are doing.

Just having 'a lot of ram' doesn't provide all the benefits that can be had with that ram if its wasted keeping a shitload of code and resources (like images) in ram that will never be used just because they are part of the processes address space.

The fact that the VM in Windows is coded in a way that EXPECTS to be able to page to disk also causes you performance issues when its turned off. Sure it can work better off, in the right circumstances, but I can assure you from your post that you don't know those circumstances or you wouldn't have suggested turning it off.

Re:When do people get this

[m_pll]

> The Win7 task manager does show a "cached" stat, though, so your effectively free memory is "free"+"cached".

Not quite. "Cached" includes pages on the modified list, which are not immediately available (or may not be available at all, if your pagefile is full/disabled... (which is one of the reasons why you shouldn't disable it, BTW))

"Effectively free" memory is shown as "Available" in win7's task manager. It is the sum of "Free" (or more precisely, free+zeroed) and "Standby". Check out the memory usage chart in the the resource monitor to get a better idea of how all these counters relate to each other.

Re:When do people get this

[Danse]

Ok, I'm truly curious here. How the heck would the OS know what memory space corresponds to the active tab in FF versus the inactive tabs or other application data?

Most recently accessed would be a good way to determine recent tabs, and most frequently accessed (or the one where the program counter is pointing) gives you the main executable. Just start by caching the least recently and frequently accessed pages first, your necessary application data to reload should be the last thing out of RAM. QED.

Isn't that essentially what it's already doing?

Re:When do people get this

[Bakkster]

Isn't that essentially what it's already doing?

It should be, hence why just using page faults and percent of occupied RAM doesn't give an accurate picture of how efficient their caching algorithm is.

Re:When do people get this

[GooberToo]

Excellent point and one I'd not previously considered.

Re:When do people get this

[DMUTPeregrine]

On Linux you still shouldn't turn your pagefile off. You should just set swappiness to a lower number, say, 1. That makes the OS pretty much avoid swapping when possible, but also prevents failures when the swapfile would be needed.

Re:When do people get this

[GooberToo]

When you memory map a file, it does do so by making it part of the paging system

I was trying not to be too technical. I did not mean to imply it copies it to the page file. Though I must admit, I don't get that impression from re-reading, but I can certainly see why you may have thought that based on the greater context, whereby that is the case.

Thanks for clarifying for others who may have thought that's what I meant.

Re:When do people get this

[Anonymous Coward]

And how all of this beats a system where applications rely more on shared libraries consuming less ram and needing not even the first disk access to swap out pages? :D

Re:When do people get this

[drinkypoo]

you miss read,

No, I'm a mister.

all that aside, due to word and dword alignments, even under compression an executable on 64 bit architecture may have to send a 64 bit wide instruction rather then 32 bit wide

I addressed both executable size and size in memory. I addressed executable size first.

so executables often will simply be larger due to padding issues

They will often simply be smaller because they can handle operations on 64 bit data types in a single instruction, resulting in fewer issued instructions.

I was suprised by the fact they weren't defacto twice the size but the amount of code that must be fixed width (and byte aligned) is less then

When?

I expected (I would assume compiler optimizations.)

Probably not. GCC has had the most optimization work on x86. I would assume large data types being handled with a simpler macro.

Re:When do people get this

[GooberToo]

I will add, should a large amount of paging be required, IIRC, it is possible for a memory mapped file to make its way into the page file. Though that's not the normal expectation. This should be known because of potential security implications. Though honestly, not likely to be an issue for the vast majority of people.

Re:When do people get this

[hairyfeet]

Good Lord man, what is this, 1998? Hell I am a total cheapskate and still manage to have more HDD space than I know what to do with! Not to mention flash sticks are dirt cheap now, so you can use Readyboost to cache to the much faster SSD. I have found with my machine, which isn't top o' the line (Phenom II X4 925, 8Gb of RAM) that having a 4Gb swap file with the same in Readyboost seems to give the fastest performance, especially from sleep.

So don't be a cheapskate dude. I'm sure you probably have at least 100Gb of free HDD space or more, is throwing a Gb or two really gonna hurt?

Re:When do people get this

[Anonymous Coward]

Best. Post. Ever.

Re:When do people get this

[Anonymous Coward]

I ran linux till last year with 512 mb with top reporting 0k of swap unless doing gimp stuff or dealing with leaky betas. I keep server and db stuff in the background too. IIRC windows 7 is at 95% OEM installed. Therefore the swapping of 88% newer PCs, which come with gigs of ram, means something is wrong. That is, 7 might have the smartest virtual memory allocation scheme of the universe but it swaps. Access to swap kills performance and responsiveness in a way that is difficult to compensate for with optimizations IMHO.
All of this is theoretical of course, win7 users could well try a linux distro and see how it performs.

Re:When do people get this

[petermgreen]

It could make sense to page out the entirety of firefox, so as to have more physical ram free for caching of game-content.
Unfortunately it's very difficult for the OS to tell whether it makes sense or not.

So you run a long batch process that access loads of files once (say an overnight virus scan) and come back to find firefox entirely swapped out and taking an age to grind it's way back in.

Re:When do people get this

[hydroponx]

Thanks for calling me a redneck and accusing me of making baseless accusations. Nice try, however I'm a white Eastern European and an Engineer.

So because you're a white Eastern European Engineer you can't be a redneck? Well, I guess that depends on your definition of redneck, however, I know plenty of rednecks from all over the world. While I'm not saying you are, Location and Profession do not determine whether or not you are a redneck, if it did then profiling would be more appropriate in all cases. You don't think profiling is a good thing do you ?

Re:When do people get this

[petermgreen]

Even that isn't ideal. It makes sense to swap out some library that hasn't been used in hours or days. The more ram available for disk cache the better.
Just because something hasn't been used for days doesn't mean it won't be used in the next few seconds.

IMO what matters most for perceived performance is having actions that should be near instant actually be near instant. If firefox has to grind it's way back into ram before it can pop up the window I want because some batch process cased the system to favour disk cache over firefox that is a fail as far as i'm concerned.

Re:When do people get this

[Anonymous Coward]

Wrong.

PAE

Re:When do people get this

[Samah]

SWAP is using the HDD as extra RAM when the system doesn't have any more memory left to use.

Assuming that the system's physical RAM is less than the addressable range for the system architecture. If you have 4GB of RAM on a 32-bit system, a pagefile becomes somewhat redundant. Once you move to 64-bit, your addressable range becomes much larger (16 exabytes), such that your motherboard wouldn't support that much physical RAM (not any time soon, at least).

Re:When do people get this

[Samah]

And just to swim in anecdotal waters, when I copied big piles of files to my GRiDPad 1910 via null modem cable using Microsoft's classic INTERLNK and INTERSVR for file sharing, SMARTDRV sped up the copy operations by about an order of magnitude. Life without disk caching isn't worth living.

Bah! All the cool kids used LAPLINK!

Re:When do people get this

[drinkypoo]

I used an official traveling software laplink serial cable, but I didn't have the software and INTERLNK works just fine without having to warez anything. When combined with xcopy /s/e/v I had all of laplink that I wanted.

Re:When do people get this

[DavidRawling]

Actually there is. In C it's malloc().

Or did you mean a callback into the running app? No, there's no requirement on Windows for an application to define an interface (function) that the OS can call. That's what paging is there for.

Re:When do people get this

[petermgreen]

The actual numbers reported for the apps added together frequently exceeds the actual total RAM, no matter which available stats you use.
That is common on many operating systems because pages can be shared between apps.

Re:When do people get this

[maxwell demon]

Actually there is. In C it's malloc().

Or did you mean a callback into the running app? No, there's no requirement on Windows for an application to define an interface (function) that the OS can call. That's what paging is there for.

I meant the latter. Especially with with garbage collected languages becoming widespread, I consider that a must. Imagine a Java program and another (non-Java) program running side by side. The Java program initially does some large allocations, which become garbage soon, but don't get collected by the Java VM because, after all, the Java program currently doesn't need the memory for anything else, and there's currently enough memory, no need to collect. Now the other program needs a larger amount of memory. But the memory is filled up with Java garbage, and the OS memory management of course has no clue that it's garbage (how could it, it's assigned to the JVM), and the JVM isn't informed that more memory is needed (that's what you said) and therefore cannot give the memory back, so the OS has no choice but to swap that memory out. In other words, it will waste a lot of time to write garbage to disk. Even worse, when the JVM later collects, it will force all that garbage into memory again (causing parts of the working set of the other program to be temporarily swapped out), just to collect it.

Or in short, the machine will slow down considerably just because it has to swap garbage out and back in, while the memory would have been sufficient if the JVM had just been informed that some program needs more memory.

Re:When do people get this

[snemarch]

Ah, thanks for the clarification - didn't know that cached included dirty pages!

As for "shouldn't disable pagefile", that really depends on how much memory you have and how heavily you push it.

Re:When do people get this

[iCEBaLM]

That's not paging.... That's just loading the file again...

Re:When do people get this

[jon3k]

"resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks"

Re:When do people get this

[FooBarWidget]

Yeah, this just proves how complex memory management is. Unfortunately this does not stop people from thinking that they're experts and concluding that software XYZ is hogging memory.

Re:When do people get this

[Josh04]

Sorry, I'm confused. Your post seems to support disabling it while the article you quote is pretty ardently against it?

Re:When do people get this

[Josh04]

A system where you're running entirely from ram *is* one with no HDD, really, isn't it?

Re:When do people get this

[mjwx]

RAM is wasted when it isn't in use. The fact that the task manager in Windows says your RAM is used 95% tells nothing, and no it won't "result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks". I'm actually really surprised, and not in a good way, that "chief technology officer" of the company doesn't know this.

But isn't the OS meant to free up this RAM for applications when it is needed?

I know this is empirical (yes, not the plural of data) but when I run 1 VM with 1 GB of RAM my computer slows to a crawl (more specifically Firefox). My PC has 4 GB of RAM and a C2D E6400. Before I upgraded (used loosely) to 7, on XP I ran a full Win 2k3 domain (DC/EXCH, App, 2 XP clients and sometimes a SQL VM) on the same machine and Firefox was completely unaffected, still browsing at the same speed as if nothing else was happening. It seems to be RAM bound based on the performance tab in task manager but I haven't gone any more in-depth then that.

Right now, with two firefox windows and outlook open the PC is using 1.32 GB of RAM. Firefox is using 241M of memory and I'm willing to bet most of that is the helpdesk system which uses a metric shitload of javascript.

Re:When do people get this

[theaveng]

>>>Only difference is, the USA is on a smear campaign against Toyota and all import vehicles.

Yes because Toyota *deserves* to be smeared:
- Engines were dying after only 20-30,000 miles, and should have been replaced under warranty, but Toyota refused. Toyota said the problem does not exist...... until the U.S. DOJ *forced* them to act.

- Same with accelerator pedals that suddenly "turn on" and refuse to stop. Toyota refused to acknowledge the problem, saying there is no problem, and several people died. AGAIN the U.S. DOJ had to *force* toyota to act.

- And so on.

Right now Toyota's actions are reminding me of American companies back in the 70s (Ford refusing to acknowledge their Pintos were exploding). Or more recently, like Microsoft towards the 40% failure rate of Xbox360s, who refuses to admit a problem exists.

Re:When do people get this

[theaveng]

No. After all you need someplace to load programs (like MS Word) from, or a place to store the documents. That's what the HDD is for.

Re:When do people get this

[Anonymous Coward]

What a crock ("RAM is wasted when it isn't in use"). Ever heard of SSD drives? Whenever you added RAM to Windows XP the machine received a nice performance boost - not so in Vista, not so in Windows 7. This over engineering is killing Windows, users should have a CHOICE of what gets cached what does.

Re:When do people get this

[Daengbo]

Why are you arguing with me? I never said 8GB was too much or that the article was well-founded. 8GB is about four times what the average computer is being sold with now, though.

Unless you are using Egacs (Eight gigs and continuously swapping, a.k.a. Eclipse), it's going to be hard to max that thing out. Sure, the OS should use as much as is available for paging. No one is saying otherwise. Swapping to disk is different, though.

If fast_turtle had said "Hey, I have an average setup with 2GB and it never hits swap," then that would have been something useful to add to the conversation.

Re:When do people get this

[Stray7Xi]

And yet when I turn off swap on my 32-bit Vista laptop performance increases 1,000 - 10,000% easily.

Dear god did you try to put your page file in the cloud?? Your box is majorly screwed up if that makes that big of difference. I'd say you either have a rootkit or your hard drive is failing. Check your SMART data.

Re:When do people get this

[Anonymous Coward]

Where do you get this type of illogical thinking from? That free RAM in the XP days INCREASED performance, now Windows "guesses" what apps you are going to use and wastes you're RAM when in fact it should be FREE and improving the speed of the foreground apps in the process. Hello?

Re:When do people get this

[Anonymous Coward]

True. Even OpenSolaris uses up as much RAM as possible by filling up the ZFS ARC which acts as a cache for data from all active storage pools.

Re:When do people get this

[drsmithy]

It pays for it because for just about everything there is a paging cost, even if its a deferred write to the page file. In an environment where paging is not typically required, it means Windows is technically running slower than need be.
[...]
But now that RAM is cheap and plentiful, in my opinion, their "optimization" now only serves to slow systems down and confuse the hell out of people who don't understand how Windows uses paging.

If pages are being written out when the system is otherwise idle (as Windows does), then there is no "slowdown".

Re:When do people get this

[Anonymous Coward]

Yes, much slower.

Re:When do people get this

[drsmithy]

Doesn't require advanced knowledge to make sense of that ouput, or understand how memory is being used.

Yes, it does. Even a "power user" is probably only going to understand "free", while Active/Inactive/Wired/Cache/Buf will just come across as static. Heck, I spent years adminning FreeBSD systems and I had to go and remind myself exactly what was what just then.

The typical end user only barely grasps the difference between the RAM and permanent storage, on a good day.

Re:When do people get this

[TheKidWho]

Location and Profession do not determine whether or not you are a redneck

Yes actually, it does.

From wikipedia:

To outsiders, it is generally a term for white people of Southern or Appalachian rural poor backgrounds -- or more loosely, rural poor to working-class people of rural extraction.

Redneck is a disparaging [1] term that refers to a person who is stereotypically Caucasian and of lower social-economic status in the United States, particularly referring to those living in rural areas.

By that definition from wikipedia then no.

But you don't make up your own definitions for words as you see fit do you?

you remind me of my redneck pal who says anybody who drives anything other than Dodge is a fool. He bases his opinion on..... well basically nothing, and I bet you do too.

Going back to your original point, no, you're wrong. Not a redneck and I tend to base my opinions on facts and physics.

Re:When do people get this

[lpq]

It doesn't say the RAM is in use for cached data.

It says the RAM is FULL...read the lines of the original post carefully -- it says the consequence of of this full RAM is requiring programs to turn to virtual memory to handle tasks....MY GAWD! you even quoted it, you twit.

The OS doesn't use virtual memory unless it is OUT of memory -- that means it's way past the point of cached data being dropped.

I pity anyone stupid enough to move to Win7 and choose to stay at 32 bits. At 64 bits, 4G is a good start system for those doing simple desktop tasks. You want to use 64 bit programs that use large address spaces? 4G is low end, and you will page.

**I** page (though rarely) with 24G (maxed until 8G Ram chips become available and cheap enough) of memory. But that's with things like Mass Effect-II and Photoshop, and a 1.5G copy of FF in memory, among many other things all in memory at the same time -- isn't that what a multitasking computer is for?

-l

Re:When do people get this

[drsmithy]

BeOS and Haiku did/do this, but I don't think any other OS has implemented total RAM usage to such a degree.

Er, what ? Windows, Linux, FreeBSD, even OS X have been doing this for over a decade.

Re:When do people get this

[complete loony]

I believe in vista & 7 if you run a memory greedy application that needs heaps of ram, causing the working set of other programs to be swapped out. When the memory is given back, the OS will start swapping that memory back in.

eg, if a batch job runs on your machine overnight, you wont need to wait for everything to swap back in when you unlock your machine in the morning.

Re:When do people get this

[hairyfeet]

Actually even on the cheaper build 3Gb seems to be settling as the "sweet spot" on everything but low end lappies, which is still at 2Gb. Even my $500 middle of the road builds I sell come with 4Gb, RAM is just so cheap throwing in that extra stick gets me so many "oooh how fast!" repeat business it is worth the $20.

But if you are just wanting to know how Windows 7 runs on 2Gb, your old pal hairy can give you the skinny. You see I have had a LOT of people that got stuck with Vista "Worst Buy specials" that come to me wanting to know if I can help them destroy the evil that is Vista. For them I insist on a 2Gb minimum, because I refuse to cripple a machine like Worst Buy does and starve it for RAM. That is why my rules are at the least 512Mb for Win2K, 756Mb-1Gb for WinXP, Vista = The Devil, and Windows 7 - 2Gb. So since many of the Worst Buy specials had a pissy 512Mb (no shit. Vista Basic on 512Mb. Talk about shitty) they usually go for the 2 1Gb sticks, since 2Gb is usually max those Worst Buy specials can hold.

The verdict? Pretty damned good actually. On 2Gb of RAM it will usually have about 1Gb free, depending on how much bling I leave on, and it is surprisingly snappy, especially compared to the bloated whale carcass that is Vista. The last was a 2.2Ghz AMD single core, and with an ultra cheapo 4350 thrown in for hardware acceleration it was actually a joy to use. Videos were smooth as butter, no skipping or glitching, really nice. I advised her to pick up a really cheap flash stick for Readyboost and with a $5 2Gb Big Lots flash stick for Readyboost it is VERY snappy, and awakens from sleep in a snap.

Now I won't try to BS you and say it don't ever hit the swap, as I'm sure that it does. It is just with Readyboost and intelligent caching I, the user, never felt any slowdowns and that is what matters to me and my customers. Windows 7 has really improved over XP, which seemed to hit the swap no matter how much RAM you had free. With Comodo AV and the PC set to autoupdate along with Firefox with ABP the PC purrs like a kitten and the customer is very happy to have the evil that was Vista flushed down the toilet. So if you want to know how Win7 runs on 2Gb, the answer is sweet, even with single core CPUs. I would just recommend with flash being so cheap (Big Lots has 4Gb for $10) using a flash stick for Readyboost, as I've found even with 8Gb of RAM having a couple of Gbs of SSD flash for Readyboost does make a difference especially in waking from sleep. And sorry if I sounded like I was coming down on you, wasn'y my intent, just pointing out with RAM so cheap the numbers are getting pretty decent now.

Re:When do people get this

[cbeaudry]

I'm not aware of Toyota not respecting warranty on engine problems. Can you please post a link? (I honestly didn't know about that.)

Pedals dont just "turn on", it sticks, in 0.0001% chance probability.

BTW, the U.S. DOJ didint do shit to Toyota, its the DOT (Department of Transportation).

There is NOTHING and I mean NOTHING different here than the dozens of recalls EACH, that FORD, GM and Chrysler (as well as others) do every single year.

Hell, in October FORD recalled 4.5 million cars because they could catch FIRE. (Remind you of the Pinto????@?!??!?!)

http://www.thaindian.com/newsportal/business/ford-recall-2009-affects-45-million-vehicles_100260633.html

These kinds of recalls, and delay tactics until its absolutely necessary (its a numbers game) are exactly the way the whole industry has been functioning for 40 years.

Why smear Toyota all of a sudden?
I explained that in my post above.

Re:When do people get this

[Anonymous Coward]

You are a whiny little fag.

Re:When do people get this

[seeker6182000]

If you disable the page file, you don't disable demand paging (guys, swaping went out of vogue eons ago). What you do is you chnage how the OS can respond to memory pressure by forcing it to write out or repurpose pages of files in memory. For example the code not currently being executed, or data files. Remember only process private pages are writen to the page file (think heap and virtual alloc). It also forces the OS to use more physical memory for things like thread stacks, which grow dynamically, and change thier physical memory usage. If you disable the page file the entire commit size of the stack (typpically 1 MB for windows apps per thread) comes from physical memory. With the page file enabled, only a few stack pages will be in physical memory and the rest are accouted for by space reserved in the page file. Since a typical system with a few applications will have between 500 and 1,000 threads you could be blowing a 0.5 to 1 GB on stack pages the system will never use. Disabling the page file is an extreemely bad idea if you really want your system to remain performant. There are cases where it can help, but most users don't know enough about how demand paging works to know when it is appropriate.

Re:When do people get this

[seeker6182000]

That is an example of demand paging. There are lots of different ways a modern memory manager pages, and what was described above is paging for those that understand how OS memory management really works.

Re:When do people get this

[Eivind]

True, the OS can't know.

It can guesstimate though. Which should I do ? Swap out this page that hasn't been accessed at all the last 2 hours, or drop this page of disk-cache that has been read 11 times, the last time 2 minutes ago.

It's a guess, but it's a REASONABLE guess that it's more clever to swap out the page.

But you're right, a drawback is that if you ain't used a program for a long time, then swapping it back in can mean a significant delay the moment you DO start using it again.

Re:When do people get this

[seeker6182000]

No that tells the OS to keep all of your process private pages in memory. Anything backed by a file will be purged from memory, like code and data files when other processes need memory. Sequential reads from code files (.dll, .exe) and data files tend to be faster than random IO from the page file. How much forcing the OS down this path impacts performance is very dependant on the memory usage patterns of the system and all of the processes executed on it.

Re:When do people get this

[yoyhed]

Exactly - it's called Superfetch in Windows 7. Win7 indeed uses a lot of RAM when it has room to spread out, preloading applications you commonly use so they launch faster. But when another app needs the RAM, it'll release it. I can confirm this - a clean install on my 6GB machine uses 1.25 - 1.5 GB, and on my 1GB netbook Windows only uses around 400MB.

I've confirmed that Win7 has excellent memory management - with a little program using an infinite loop of fork() and malloc(rand()). On my 6GB machine, it quickly spawned over a thousand processes, using up 98% of the memory. However, as more and more processes spawned, memory usage stayed at 98%, because Windows kept releasing more and more of its own memory to the program as it requested it. The system was still responsive 10 minutes later (I don't remember how many processes were running by then). Once I killed all the processes with taskkill, Windows was only using about 200MB - about 100MB more than XP uses at a minimum.

This CTO needs to get with the program - this is 2010, and Windows has been using prefetching technology to users' advantage for 3 years now.

Re:When do people get this

[BikeHelmet]

RAM is wasted when it isn't in use.

This is a common line fed by the media to anyone that will believe it.

RAM does have a limited amount of bandwidth. Dumping all those memory pages (and zeroing them) does take a measurable amount of bandwidth. How much bandwidth does your RAM have - 25.6GB/sec? Zeroing out 2GB to use it could potentially take up to ~100 miliseconds. Definitely measurable.

I will agree that memory is wasted when it isn't used - but we need mechanisms in place to direct programs in how they should behave.

Got two HDDs? Started defrag on one? It'll use up a TON of RAM (1+ GB) to defrag a bit faster, when it could probably get by just fine with 30MB. Now start a game. What's that you hear? That's your HDD swapping.

If programs could constantly register and update the amount of low-priority and normal-priority freeable ram they have with a daemon, that daemon could invoke freemem calls for apps based on the amount a program needs, and the amount other programs claim to be able to free.

A paint program might advertise memory for all the plugins it uses, and any open images, but only when not in focus.

A game might advertise the amount of memory it uses for textures that weren't used in the past few frames, and perhaps map or level data that isn't visible. Obviously this data should have a higher priority, because reloading it during play could cause stutters or framerate drop. HDD IO could be another factor in priority. If data can be re-generated without HDD IO, that memory is prioritized higher when some needs to be freed.

The daemon could have access to CPU usage stats, ensuring active programs never get tripped up by other active programs, as long as an idle program is available.

A web browser might advertise all hidden tabs, thus allowing games or other high demand apps to usurp hundreds of megabytes. When you alt+tab and the game frees its memory, the web browser could automatically start regenerating stuff in RAM, so that when you flip tabs the next page pops right up.

Until we provide the tools to developers, nobody will be able to envision all the uses, or how to do it properly. But we do know that doing nothing isn't working, since every new OS release it's the same thing all over again. Something like this could grow to vastly improve memory usage, but only if we let programmers evolve it. The kernel is not smart enough to page only unimportant stuff to the HDD with 100% accuracy. We need to let programs help out, by offering them the chance to literally dump data that isn't deemed necessary, when necessary. In the process we save HDD IO - done properly it could be a huge speedup.

P.S. Linux has some sort of kernel memory compression that could boost memory capacity by about 60%. The Pandora devs were eying it, since it'd push 256MB to about 400MB effective. I have a quad-core, so memory compression probably wouldn't hurt, especially if it was smart about compressing memory from idle processes. It's got to be faster than paging to HDD.

Wow, that was quite a well of text. I hope I articulated that well enough.

Re:When do people get this

[Waccoon]

Some programs require a swap file. Photoshop, for example, won't work without it, and will flat out tell you to turn the swap file back on. This, despite the fact that Photoshop uses its own proprietary swap file.

I'd imagine that there are many other apps that will enforce the use of a swap file, no matter how much memory you have.

Re:When do people get this

[AVee]

Ok, I'm truly curious here. How the heck would the OS know what memory space corresponds to the active tab in FF versus the inactive tabs or other application data?

It wouldn't know about tabs as such. However, it could track which page was accessed/changed/created most recently. Memory related to the currently active tab is likely to be the most recently used memory.

Re:When do people get this

[ultranova]

I'm assuming you're joking, since I said nothing about amperage, and naturally if we're talking about an actual flip-flop unit in a stick of RAM, we're going to be talking about mA or even nA.

Which would take the peak power requirements down to 200 megawatts or 200 kilowatts, respectively. That's not exactly realistic either, if you think about it.

But yes, I'm joking.

Re:When do people get this

[sjwt]

It depends what they are looking at, the Graph doesn't show cache, but the stats do. i.e. my whopping 2GB lists 319meg free, but the graph shows only 41%(842meg) used, looking deeper into the resource monitor, it too shows only 840ish used megs, but also lists 820megs in 'standby'

This is on Windows7 Home 64bit

Re:When do people get this

[Josh04]

And when your program is large enough that all the things it needs can no longer simply fit in RAM? What then?

Re:When do people get this

[Anonymous Coward]

This isn't actually true. If the whole of physical RAM is filled up with disk cache, and you launch a new program, then the OS has to look through the allocated memory and find something to either drop of shuffle to the swap file, using a complex algorithm. This can cause slowdowns. As such, you will see better performance if you set the disk cache to always leave some RAM unused, and set the swapping algorithm to start swapping to disk before RAM is completely full. Of course, Windows doesn't do that - it does what you suggested, which is sub-optimal.

Re:When do people get this

[herve_masson]

It seems that most people who complain about memory usage don't know how memory is managed on modern operating systems, so they go all apeshit about "OMG HELP linux is using so much memory it sux0rz!!!"

I agree with you on that, but feel like the diagnostic is somehow ... opposite: "it seems that linux developers fail to show memory usage in a way that average user can understand".

Re:When do people get this

[Anonymous Coward]

You mean the one about "lies, damn lies and statistics"?

Re:When do people get this

[Man Eating Duck]

I'm for keeping a pagefile :)
If my post was unclear, I experienced an increased loading time (which is bad) when I disabled the pagefile. I believe this is because Windows couldn't use that memory for disk caching because more of it was occupied by the game I was playing.

Re:When do people get this

[Foolhardy]

I don't know that it's documented in detail anywhere. The beginning priority seems to be based on the process's priority (5 for normal apparently) and is adjusted by usage heuristics and superfetch. There is a overview here. It may help to raise the priority of FF to AboveNormal if it seems like its pages are being discarded unnecessarily.

What the other posters said in reply to your other post about the OS not really knowing what memory belongs to what tab, instead having a page level view of things, is correct. When the CPU accesses a page, it sets a flag in the page descriptor that the page was accessed. The memory manager checks these flags periodically to see what pages are being used. When the MM thinks the process has too many pages, it takes away those that haven't had that flag set in a while. I guess the frequency of usage has some effect on the priority, but I'm not sure.

Re:When do people get this

[ShadowRangerRIT]

The point isn't to save HD space, it's to reduce paging activity. HD space is infinite, but HD read/write capacity isn't, so if you honestly don't trust the paging algorithm, you want to keep it small so it doesn't overdo it.

Re:When do people get this

[GuyverDH]

Not if you've disabled the pagefile altogether.
Don't forget to disable the paging executive as well.
If running Windows XP, see the following..
http://kadaitcha.cx/performance.html

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management]
DisablePagingExecutive - set to 1.

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management]
LargeSystemCache - set to 1.

Re:When do people get this

[Josh04]

Ah, my fault, I read "increased speed in loading times" XD

Re:When do people get this

[Anonymous Coward]

You have it backwards. Pages Output/sec is pages written to pagefile - not a big deal. Pages Input/sec is pages retrieved from pagefile (or other backing store) because they could not be found in memory.

Occasional spikes of pages input are unavoidable (for instance, reading more from an executable that was originally demanded), but the goal of the operating system is to page-in as rarely as possible. When you see a system that's constantly paging in, you're seeing a system that's short of available memory and is thrashing the disk.

So pages input/sec absolutely *is* an indication of running out of memory.

Re:When do people get this

[jhfry]

I have never used a Video Editing or Photo Editing package that didn't require that you specify a directory for it to use as it's cache. I think those kinds of apps know that even a pagefile is often not adequate when working with the volumes of data in question and they don't really even use them (except where the OS does it for em).

Re:When do people get this

[Anonymous Coward]

Stuff does not slow down due to that swap out. Vista/7/2008 have IO prioritization. Page-outs (from RAM to disk) have lowest priority, so basically anything else that wants to read or write the disk goes to the front of the line.

Re:When do people get this

[iCEBaLM]

Demand paging is the act of only loading the parts of an application that is needed at the time requested. If you've already loaded a file into memory because an application needed it but then freed that memory to make room for other stuff on the assumption that you would just load the whole file again when needed then that is not demand paging. That is just loading the file again. That is also an attack vector and poor memory management.

Re:When do people get this

[seeker6182000]

Please show where I said the whole file was loaded. Oh, that's right you can't. Windows uses demand paging, and if you turn of the page file, it works like it should and not what you claim I said.

Re:When do people get this

[Anonymous Coward]

Hmm... I never really paid much attention. But, looking at my Windows 7 box: 3233 MB In Use (about 95 apps + services); 73 MB modified (active use memory); 9003 Standby (Cached Files); 1 MB Free... 99.9%+ Used, and I'm quite happy with it! It's about damned time Windows uses my 12 GB memory efficiently!

The extra 8 gigs *hugely* supports my video/photo editing/converting, but I'm ecstatic to see it improving my day-to-day workflow opening and closing commonly used apps when not in use!

Re:When do people get this

[FooBarWidget]

Why "Linux developers" exactly? If you look at Windows, OS X, FreeBSD, or pretty much any other operating system, they all report memory usage the same way: VM size, RSS, etc. None of them display memory usage in a way that average users understand. I don't think there's even a way to do that, it's like blaming airplane designers for not designing an airplane that average people can operate.

The simple mental model of memory usage that people have is not only wrong, it is also vaguely defined. Virtual memory, along with disk caching, have many optimizations and are complex by nature; there's simply no way to reduce that information to a simple mental model and be 100% accurate at the same time. There are only two ways to display memory usage in a way that average people understand:

- By not using virtual memory and not using disk caching, i.e. by making your computer inefficient, dog slow and prone to crashes.
- By only displaying "guesses" of what the "real" memory usage is, for some vague definition of "real".

Actually when it comes to the latter, Linux has an edge here over other operating systems. Recent kernel versions introduced the concept of "Proportional Set Size", which is something like the RSS but it divides the size of each memory page by the number of processes that share it. This is as close as you can come to what people expect to be the "real" memory usage of processes. The standard OS tools don't display this number yet but I'm already incorporating this feature into my own tools.

Re:When do people get this

[iCEBaLM]

You didn't, the OP which you are referring to did, and you agreed with him.

Re:When do people get this

[zippthorne]

That workaround was good, but not nearly enough better than just setting the max and min size of the swap file to the same number and defragging to justify repartitioning. And if you had enough memory, it was still better to just turn it off entirely.

Re:When do people get this

[Anonymous Coward]

Bad advice. I decided to do that eons ago. Upgraded my video card to a more high end gaming card and suddenly I started having all sorts of problems with some games not showing the higher textures or just completely crashing randomly. 4GB of system ram meant that I could more than hold the required information. However for some reason if you do not have a page file, some games at higher resolutions/texture quality behave oddly.

Re:When do people get this

[ozmanjusri]

Did I miss something? Since when was defending Windows the hip and trendy thing to do on Slashdot?

It started a few years ago.

Microsoft realised just ho much bad press was being generated by tech and social networking sites, so they employed an advertising agency called Crispin Porter + Bogusky. They arranged online reputation management for Microsoft, including a "Downmod Squad".

Re:When do people get this

[GravityStar]

Why do people always bring this up? And it's marked insightful, like it's some ingenious find in a pool of mud. While PAE technically gives you the ability to use more than 4GB of address space on 32-bit x86, it's irrelevant to the average user and also to nearly every power user.

Enabling PAE and also actually making real use of it will require a application that is built for it and tested for it.

If we consider the windows software 'ecosystem', and exclude video-editing software and server software: How many applications will actually work with PAE? My guess the answer will be nearly nil.

And still, people bring up PAE as some sort of solution for the ordinary computer user. Non-techies have come up to me and _told_ me there was a way for a application to use more than 4GB of ram in 32-bit windows. And they where grasping at straws, thinking "If only I could get that to work, stupid techie telling me I'd best not do that and just upgrade to 64-bit instead... Wait, my brother's niece's kid is pretty handy with computers, maybe I'll ask him instead".

Re:When do people get this

[AzN1337c0d3r]

Wrong. When you use (otherwised unused RAM as disk cache), you have 4 situations: 1) You open an app that is cached. This is massive speedThis takes virtually no time.up compared to accessing disk. 2) You open an app that is not cached. The OS clears out some room in the RAM (almost instantaneous), and loads from disk. This is no slower than the case where you had no cache. 3) An already running program requests more memory. Disk cache containing programs that are not running can be safely discarded to make room for the program requesting additional memory. This takes virtually no time. If you do not have enough room in your disk cache and have to go to disk to get memory, you do not have enough physical memory anyways. 4) An already running program discards its reserved memory. The OS repopulates the memory with disk cache data while the disk is idle. This again has no negative effect on performance. As you can see, disk cache (aka SuperFetch) can provide only performance increases and it also fills the RAM up with "crap". The Windows 7 OS itself without any of its caching technologies don't take up that much RAM, so your performance issues lie elsewhere in the OS.

Re:When do people get this

[aiht]

If it's coming from CD: decompression is faster than seeking and/or reading.
If it's coming from a cached-in-RAM copy, then I dare say the decompression would be slower than the reading.

Re:When do people get this

[ByOhTek]

Not when being read from the ramdisk it isn't coming off of the CD. Seriously, try to keep up here.

Re:When do people get this

[DaVince21]

I know when I run Puppy Linux, which fits entirely within ~30 megabytes of RAM, the system is the most responsive OS I've ever seen. No HDD caching means no slowdown or pregnant pauses.

Yeah, and modern heavyweight systems do the same: cache as much of the important OS stuff in memory as possible. Sure makes them faster, wouldn't you think?

Re:When do people get this

[DaVince21]

Um... Linux and now Windows 7? It's called memory caching.

Re:When do people get this

[Anonymous Coward]

"Where do you get this type of illogical thinking from?" says the person who thinks that free RAM merely existing causes programs that aren't using it to run faster. What amusement.

Re:When do people get this

[nmg196]

Personally, I think people rarely multi-task. Using two mice with two cursors is pretty much impossible and is not actually supported by the operating system anyway, so people tend to just use one application at once and then switch between them (which is not really multitasking).

Your last sentence is funny. Firstly, it's using 95% precisely to increase the chances of cache-hits. Secondly, you make it sound like "dumping memory" takes time. Windows 7 can free up a gig of disk cache in less than one hard disk seek time! It doesn't have to write it to the hard disk - it just instantly and dynamically marks it as free as your real programs request more RAM. So if it's not slowing down your system and has no adverse side effects, then what's the problem with it? Currently everyone I see complaining about it doesn't actually seem to understand it (or memory management in general).

Re:When do people get this

[makomk]

On the other hand, hard disk access isn't going to be any faster on an Athlon X2 than a P4, and Windows Vista adds new sources of hard disk IO (SuperFetch and indexing, for a start) that slow down other disk accesses.

Re:When do people get this

[clone53421]

Secondly, you make it sound like "dumping memory" takes time.

Well, what I actually said was dumping it to the swap file takes time. Which it does. Of course cached stuff can be freed nearly instantaneously if it doesn’t need to be swapped back to disk.

But yeah, I’ll admit that simply because the system is using 95% of the RAM isn’t enough to conclude that it’ll run slowly, since a lot of the cached stuff can simply be freed without having to swap it out.

Battery Problem Explanation?

[eldavojohn]

If these claims are true, isn't it possible that this could be seen to the user as a source for the battery life problems? I suppose that disk-based virtual memory would incur a little more read/write on your hard disk as well ... possibly decreasing the mean time to failure for Windows 7 users.

Re:Battery Problem Explanation?

[Anonymous Coward]

i think the battery life problems are related to some obscure task that seems to run of its own accord, even from a fresh vanilla installation... there's something called the "idle process" which I've often observed eating 90-99% cpu on a lot of machines that i've checked.

Re:Battery Problem Explanation?

[Anonymous Coward]

Users with multi-core systems might try to set the affinity for the idle process so it only chews up one core ;-)

Re:Battery Problem Explanation?

[Anonymous Coward]

The idle process is how it displays unused CPU in the task manager. You'll find that if you add up the CPU usage of every process in the list it always comes out to 100% thanks to that.

Re:Battery Problem Explanation?

[ThaReetLad]

You might want to get your sarcasm detector serviced. It appears to be mis-calibrated.

Craig Barth?

[BadAnalogyGuy]

I don't know.

Depends on what kind of memory

[A beautiful mind]

If it is filesystem cache, then it's not wasted or "maxed out". If it is application/system memory, then it is indeed a problem.

Re:Depends on what kind of memory

[Bazer]

Would a filesystem cache cause the system to swap?

Re:Depends on what kind of memory

[dhavleak]

Not necessarily -- it could be application / system memory that's pre-cached (based on profiling). If at any point, your machine has RAM available, and idle cycles, pre-caching would be a good way to use them. See here.

Re:Depends on what kind of memory

[A beautiful mind]

It's Windows. It might be a bug.

Re:Depends on what kind of memory

[Spad]

No, but they don't offer any actual evidence that the systems are swapping a lot, just that their memory usage is high. The "and this causes swapping" part of the claim just seems to be an assumption rather than a fact.

Re:Depends on what kind of memory

[clone53421]

A pretty safe assumption, since your memory use is going to fluctuate quite a bit as you use the computer.

You just loaded a website with a Flash video? Whoops, time to load Flash player. It’s probably cached, because you probably used it five minutes ago, and it’s probably been shoved to VM because you stopped using it and other things needed the memory.

Any time you start another application, it’s going to have to start moving memory to VM to make room for the application.

Re:Depends on what kind of memory

[Jetrel]

Or a feature!?!?

Read the rest of the comments.. Think and then mod appropriately.

Bogus Story

[filesiteguy]

Let's start from the story (which I *did* read) - 'Barth acknowledged that XPnet's data couldn't determine whether the memory usage was by the operating system itself, or an increased number of applications,"'

Right there I'd be suspect whether this is even an issue or not. Given Windows (which I generally regard as inferior) as an OS having lots of functionality, I wouldn't be suprised if it takes up all available RAM prior to utilizing swap. I'm on my 2GB Ubuntu system right now and am running at 18% of 2GB with just Mozilla (with two tabs) and Thunderbird. But there's also my network layer (Network Monitor), KTorrent, and my bluetooth daemon running in the background. All told, System Monitor says i have 31 processes running.

Let's do a like comparison - run the exact number of apps and processes before declaring a memory leak.

Sheesh!

Re:Bogus Story

[IchigoKyger]

Well I'm running Windows 7, in which I've never had a problem with, and running around 59 processes. Which are comprised of AVG, Google Chrome, Curse Client, Google Desktop, Intuit Update Service, Net Drive, Skype, Winamp, and WoWHead Client. With all that running I'm sitting at 33-34% Physical Memory usage of the 2GB of Memory available.

Re:Bogus Story

[LazyBoot]

I'm sitting here on win7 and can't really see the problem TFA is writing about... According to the task manager, I got 68 processes running, and using about 39% of my memory (reported at 3327MB) Yet only about 200 MB is listed as free, and the rest (about 1800 MB) is listed as cached (as it should imo)

Re:Bogus Story

[Anonymous Coward]

windows is inferior, says the linux newb running ubuntu

Re:Bogus Story

[ShatteredArm]

Agreed. Summary (and probably TFA, I didn't read it) is a bunch of FUD. I run Win7 on a VM, which is also running SQL Server and Visual Studio 2008, and I don't have any performance problems, despite allocating only 1 gig of RAM to the VM. There's no way in hell Vista can do that.

Re:Bogus Story

[Anonymous Coward]

There are a BUNCH of different kinds of memory in windows. Which all map back to physical memory.

The easy way is to open task manager click on the performance tab and look at 'total' that is the physical ram installed. Then there will be a bar to the left showing how much memory is currently commited. That is the total amount of memory required by all the applications running. Then there is a 3rd type of memory. This does not show other than 1 place. It is under system cache/cached. Then there is free memory. With vista that number will be close to 50-0. With 7 I do not have any metrics yet as I have not used it much.

Also clean simple install of XP will use ~130meg, vista nearly ~700 (you can get down to 500 if you disable a bunch of stuff), with 7 I have seen about ~350-400 (out of box). Then if you have an OEM computer there is probably a bunch more junk. The quicklaunch bar from HP (the thing that controls those cool touch buttons on some of the higher end laptops) uses nearly 700meg virtual and anywhere from 40-200 meg physical. MY HP vista laptop out of box used nearly 1.5 gig of physical just to start. That was before *I* launched any of my applications. If I had 2 gig of memory I am almost 85% full. I bought 4 because I knew this going in. I have whittled it down to ~900 meg.

There are windows APIs for all of this. It will slice and dice it in so many ways your head will spin.

Honestly it is feature bloat from the OEMs that is really squeezing memory. HP should be pushing back on their suppliers to say 'make your packages smaller and more faster features'. Instead they are pushing with 'add more features!'. Then MS shows up and nearly triples the amount needed just for the OS. Really it is a case of what Intel giveth the OEM taketh away.

Re:Bogus Story

[ComfortablyAmbiguous]

I'll second that. Currently I have 63 processes running consuming 3.05 GBs ( out of 12). Another 8.8 GBs is cached, leaving me 379 MBs free. About what it should look like.

Win7 Reserves Memory

[Anonymous Coward]

Win7 reserves memory which makes it look like there's no available system memory even though the OS has plenty to share with programs once the need arises. Exchange 2007 and 2010 work the same way... they reserve all available system memory and then release it as other processes call for it. I think that if this company's CTO does not understand this concept it shows something quite negative about who they select for employees. I'm also guessing the dude's a mac fanboy. ;-)

Ah, Slashdot

[Anonymous Coward]

1. "CWMike" (Computer World Mike) submits Windows bashing link (to Computer World) on Slashdot
2. Slashdot "editor" blindly shoves it on the front page
3. Page hits to CW from /.
4. CW profits!
5. Mind numbing /. blather to follow.

It's called SuperFetch

[dhavleak]

I guess Devil Mountain or whoever don't know about SuperFetch. Or need publicity.

And I guess slashdot editors don't know about SuperFetch. Or maybe an article like this gets them more traffic, revenue, etc.

The fucking bullshit that passes for articles these days..

Re:It's called SuperFetch

[mystikkman]

True, the technical details for interested geeks are here http://blogs.technet.com/askperf/archive/2007/03/29/windows-vista-superfetch-readyboost.aspx

Trollworthy rating for posting this...

[hitech69]

Computerworld should just close up shop for this worthless piece of journalism, or at least give their author the boot for doing any work with Craig Barth who represents a team of morons. samzenpus should be given a troll rating for getting this to Slashdot.

Re:Trollworthy rating for posting this...

[Anonymous Coward]

FWIW InfoWorld blogger Randall C. Kennedy is Devil Mountain Software's "director," "president," or "CTO," depending which article you read.

Page Faults

[Dr_Barnowl]

The metric to count is the number of page faults, an indicator of the number of times that the OS addresses memory that isn't in RAM.

As others point out, measuring just the fraction of memory consumption is stupid. I have 6GB of RAM ; my processes are using about 1.7GB of that, but the OS is claiming that 3.8GB is consumed. So that's 2.1GB of cached data that I no longer have to wait for from disk. Hooray.

TFA hints that they may be measuring page faults, and does mention that Win7 is hitting the disk for virtual memory more often. But they should make that clearer if it's the case.

Re:Page Faults

[DJoffe]

Hmm, it might be instructive to clear out our terminology here, to avoid confusion ... "page faults" as the term is used in Windows are a useless metric in terms of speed, as that doesn't tell you how often the system is hitting the disk. When an application allocates memory (e.g. 'char* foo = new char[65536];) the system doesn't actually allocate any memory at all until the application attempts to read from or write to those allocated pages (similarly the terminology here also gets confused as this is also called 'virtual memory' while others use the term 'virtual memory' to refer to hard disk swapping specifically) ... in Windows terminology a "page fault" is counted when this happens ... that's just normal and harmless and doesn't necessarily have anything to do with the hard disk at that point. According to Wikipedia, "Note that Microsoft uses the term hard fault in its Resource Monitor to mean 'page fault' (cf. Resource View Help in Microsoft operating systems)". If you say "The metric to count is the number of page faults", I presume you mean "hard fault" (i.e. memory in disk) rather than simply memory that hasn't been allocated at all.

Re:Page Faults

[greg1104]

I'm not sure they're just measuring memory consumption. It looks like the metric used may be what is described by the vendor as:

"The Peak Memory Pressure Index is calculated by comparing a series of 4 independent Windows metrics – the Memory\Committed Bytes counter, the Memory\Pages Input/sec counter, the PageFile\% Usage counter and the aforementioned event duration value – against a set of user-defined threshold values. The resulting individual ratios are then weighted and combined to create the single number Peak Memory Pressure Index value."

That seems equally bogus and needlessly complicated to boot though; I wouldn't want to be the guy who had to defend that measurement as useful.

Uhn...no.

[Pojut]

People are either being unbelievably stupid, have only 1 gig of RAM installed, or this is FUD. Example: I'm currently running Windows 7 64 bit. On my secondary monitor, I have a bunch of system monitoring widgets... hard drive space, CPU load and temp, video card load and temp, memory usage, etc. Just last night I was playing Bioshock 2, all settings at max. Even with those widgets running, with Aqua Teen Hunger Force playing in MPC on the secondary monitor in a window, and Bioshock 2 running full bore, I was still using only 52-58% of my available 4 gigs of ram. I call BS on this article.

Here are my system specs, to back up my claims. As you can see, nothing special (copy and pasted from my [H]ard|Forum sig):

Display: Asus VH236H | Dell 2005FPW
Foundation: Cooler Master Storm Scout | OCZ ModXStream Pro 700w
System: Gigabyte GA-MA785GM | AMD Athlon 64 X2 5400+ | Corsair XMS2 4GB DDR2 800 | ATI 4850
Internal Storage: Diamondmax 21 system | WD15EADS archives
External Storage: 1.25TB in a KINGWIN DK-32U-S | WDMER1600TN
Input: Kensington 64325 Expert Mouse | Saitek Eclipse II | M-Audio Axiom 25
Audio: Logitech Z4 2.1 | Audio Technica ATH-AD700

Re:Uhn...no.

[Cowclops]

The plural of anecdote is not data... but I still have to throw in my own 2 cents. It seems that as long as you have at least 1.5GB ram, win7 will be using about 700MB of that on startup, and the rest goes to apps. I have 4GB ram and I've never seen more than 3GB of ram in use, including running Supreme Commander with other stuff open.

I agree that Win7 can't run on systems with less than 1GB ram, and only runs "alright" on systems with exactly 1GB ram, but if its using ram for something other than disk cache, then its not the OS itself using the ram, its the applications they're running.

Re:Uhn...no.

[1s44c]

People are either being unbelievably stupid, have only 1 gig of RAM installed, or this is FUD.

Just because you didn't hit a bug isn't proof that it doesn't exist. It's likely that something as big as windows 7 has various bugs that only show up under certain conditions.

Re:Uhn...no.

[ThaReetLad]

Yeah it can. It works pretty well on a VM with on 520mb of RAM, albeit with no aero. In fact the VM with 7 on almost performs as well as the XP host OS with 2Gb

Re:Uhn...no.

[Pojut]

I know a LOT of technically competent people who use Windows 7 on their gaming system (with some of them using Windows 7 on their main system, although most of us run either Hackintoshes or some flavor of Linux for our primary setup) and none of them have encountered this issue.

I'm aware that the standard "people I know" subsection isn't the same as "everyone", but we're talking 30+ people here. If this problem is big enough to affect as many people as the article claims, most of the "people I know" would have encountered this problem at one time or another...and none of them have.

For what it's worth, almost every non-tech person in my family currently uses Windows 7 as well...and again, none of them have encountered the problems mentioned in this article.

Re:Uhn...no.

[bemymonkey]

Windows 7 seems to place a lot of data in swap, even when gobs of RAM are available. With a large enough page file and Superfetch+Prefetching enabled, physical RAM that's actually in use (not cache!) is usually between 50% and 60% full.

Disable the page file, and the RAM that's in use skyrockets to 70%+ with the same processes running.

Since I have a tablet with a dead-slow 4200RPM 1.8" hard drive, disabling the page file and Superfetch gives me a huge performance boost (Firefox scrolls more smoothly, programs start more quickly), which is what leads me to believe that Windows 7 fills up the page file with active program data in order to keep RAM free. However, with this setup I regularly get Out of Memory warnings when I've got a few browser windows open and Photoshop + Excel + Word - which is a pretty common scenario.

In this case, memory usage (not including cache!) runs up to 85-90%... 2GB is NOT enough for a Windows 7 system unless you have a nice fast hard drive for the page file (the WD 5400RPM 2.5" SATA drive in my desktop replacement does fine for this, but still has a performance hit compared to running without a page file). 4GB without a page file is the sweet spot right now, but sadly, my tablet maxes out at 2GB.

As for all the idiots saying the RAM isn't actually full and is actually just being used as prefetch: The Windows 7 task manager shows how much RAM is AVAILABLE. This is cached + free, meaning the total RAM available for use with programs... and it's very much possible to hit 2-4GB of actual FULL RAM with just a few simple office programs and a web browser.

Hell, I'm using 1.3GB right now and all I've got open is Firefox with 3 tabs, Winamp and MS OneNote 2010... Win7 is a pig.

Re:Uhn...no.

[Pojut]

Out of curiosity, why are you running Windows 7 on a system that doesn't handle it well? Why not put XP or some flavor of Linux on there? I know Windows 7 has a much more "tablet friendly" GUI, but there are mods you can make to XP to do the same thing...

Re:Uhn...no.

[bemymonkey]

It runs fine if you're careful and don't turn on the page file. Once programs are open they're just as snappy as on my main systems... you just have to be careful about how many programs you open at once - especially memory hogs like Photoshop or Firefox. :)

Re:Uhn...no.

[bemymonkey]

Oh, and to answer your actual question: It's mainly because of the tablet features and indexed search :)

Re:Uhn...no.

[Anonymous Coward]

Congratulations on failing basic math. 100% - 86% leaves 14% of people who will not experience this - not just you, but another 13 buddies out of every 100 people. Learn basic stats before you go around calling BS - otherwise you look like an idiot.

Re:Uhn...no.

[Pojut]

not just you, but another 13 buddies out of every 100 people

And if you go back and read my post, you will see that I mentioned 30+ people in my own circle as never having this problem. Last time I checked, 30+ is more than 13.

Learn basic stats before you go around calling BS - otherwise you look like an idiot

Learn basic reading comprehension before you go around reprimanding people - otherwise you look like an idiot.

Re:Uhn...no.

[Pojut]

Sorry about that, not my OP...I meant this post, which is in the same thread.

Re:Uhn...no.

[Pojut]

Oh, and to answer your actual question: It's mainly because of the tablet features and indexed search :)

That's pretty much what I figured...what tablet are you running? I've considered picking up an Asus EEE T91MT for use as an internet browser/ebook and ecomic reader.

Re:Uhn...no.

[bemymonkey]

I'm using an IBM Thinkpad X41T. Bit faster than the EeePC T91, and the Wacom digitizer on the X41T is far more accurate than anything available on the EeePC. I'd recommend looking for a used Thinkpad (or possibly HP) tablet if you're planning on actually working with the device (i.e. writing/drawing with the stylus)...

Although you actually sound like you want to use your fingers for navigation, and not a stylus... which is not something I'd buy a tablet for. Windows is still heavily optimized for a mouse and keyboard, and navigating as well as text entry via the touchscreen are more of a last resort, IMO.

It's definitely something you should try out before shelling out the dough.

Re:Uhn...no.

[Pojut]

The only reason why I want the tablet functionality is for eBooks and eComics...otherwise, I would most likely use it in normal laptop mode. This is one of those "if I somehow find $500 that I didn't know I had, this is what I would spend it on" kind of things :-)

Re:Uhn...no.

[Bobfrankly1]

People are either being unbelievably stupid, have only 1 gig of RAM installed, or this is FUD. Example: I'm currently running Windows 7 64 bit. On my secondary monitor, I have a bunch of system monitoring widgets... hard drive space, CPU load and temp, video card load and temp, memory usage, etc. Just last night I was playing Bioshock 2, all settings at max. Even with those widgets running, with Aqua Teen Hunger Force playing in MPC on the secondary monitor in a window, and Bioshock 2 running full bore, I was still using only 52-58% of my available 4 gigs of ram. I call BS on this article.

Here are my system specs, to back up my claims. As you can see, nothing special (copy and pasted from my [H]ard|Forum sig):

Display: Asus VH236H | Dell 2005FPW Foundation: Cooler Master Storm Scout | OCZ ModXStream Pro 700w System: Gigabyte GA-MA785GM | AMD Athlon 64 X2 5400+ | Corsair XMS2 4GB DDR2 800 | ATI 4850 Internal Storage: Diamondmax 21 system | WD15EADS archives External Storage: 1.25TB in a KINGWIN DK-32U-S | WDMER1600TN Input: Kensington 64325 Expert Mouse | Saitek Eclipse II | M-Audio Axiom 25 Audio: Logitech Z4 2.1 | Audio Technica ATH-AD700

This, ladies and gentleman, is a nerd. He will take any opportunity to post his home system specs. Never mind the fact that TFA does a good job of contradicting itself.

Re:Uhn...no.

[Pojut]

This, ladies and gentleman, is a nerd. He will take any opportunity to post his home system specs

Even if those specs are unimpressive!

Re:Uhn...no.

[bemymonkey]

Hmm, what exactly is so great about having a touchscreen for comics and ebooks? Or is it more about the form factor?

Re:Uhn...no.

[Pojut]

It's the form factor...the touchscreen would be nice just because I wouldn't have a stylus to keep track of, but it isn't a requirement.

My experience

[Spad]

I'm running Windows 7 x64 with 4Gb of RAM; currently I'm running Outlook, Firefox, IE, Excel, FeedDemon, Office Communicator, AV, AD management tools, call management software, a couple of powershell instances, Context, RDTabs, Putty and the usual assortment of drivers, plugins and background apps. I'm at 2.4Gb of RAM; even on a 2Gb machine it would be usable, though I'd probably have to be a bit more zealous with closing unused apps to avoid swapping.

I can only assume that it's the usual nonsense of vendors shipping Win 7 machines with only 1Gb or 2Gb of RAM, loading them with crapware, putting cheap hard disks in, telling the users they can multitask their asses off and then acting surprised when the performance isn't up to much.

Re:My experience

[3seas]

your using the wrong apps and data.

You can take small efficient application as was used to be created because ram was to expensive and run hundreds if not thousands of them without a slow down, especially if teh data they are dealing with is near empty. Or you can take one autocad application and large cad file and tax it to its limits and beyond.

I'm sure there are other such applications that are massive and resource hungryas well as data files.

Increased speed and memory of hardware is generally preceived as more for the software developers to use, not for the benefit of improved usability for the user.

I keep saying that someday they will make hardware and software that can keep up with mi thr33 fingre tpuying.You kmnow which three fingers too.

Re:My experience

[Anonymous Coward]

I've gotta agree here, I hate MS as much as the next slashdotter, but Windows 7 really actually is a decent step forward.. obviously it's still not going to come close to Linux in terms of performance or stability, but I seem to constantly be seeing stories about how its actually no better than XP or Vista.

Personally I have a machine with 8GB of RAM, with the pagefile disabled, and the majority of the time, its only about 20% in use, going up to 30% at times like now, when I'm ripping a DVD, doing an x264 encode, and watching a movie simultaneously.. Only once have I noticed it hit 40% or above, when I ran into a memory leak that sent it through the roof - around the time it got to 98% I rebooted, but that was the only time that happened.

My 2c is that Windows 7 is a damn sight better than XP or Vista - I found XP unstable as all hell on my system, and while Vista was more stable, it was also fairly bloated, although admittedly I switched to using Server 2008 fairly shortly after it's launch, I'm sure it was a lot better after a few patches and the SP.. Due to my hardware configuration (fairly heavy OC, SLI, etc.) there probably more points of failure than your average PC, but since I switched to Windows 7 around the time the RC came out, I've probably had less than 10 crashes, definitely less than 20, and this is a computer I use 12 hours a day at the minimum, and a damn sight more than that on average, and which is running x264 encodes over the night at least 1 night a week.

For the record, my hardware:
Intel Core 2 Quad Q9450 2.66Ghz, @ 3.6Ghz on air
8GB Corsair DDR3 1800
XFX Nforce 790i Ultra
2x XFX Geforce 8800GTS 512MB's in SLI
150GB WD Raptor 10,000rpm HDD

People are buying cheap PC's and upgrading old one

[Nzimmer911]

"Current generation hardware"? Seriously, how many machines in this very small sample set are using i series intel chips? The way windows 7 was marketed, I'd bet that many of these machines were upgraded XP boxes. Top that with the 32 bit memory caps and people's general hesitation to install a 64 bit desktop OS, and I am not surprised at all that many machines are hitting memory saturation. Add to that that the Windows 7 interface leads to leaving more apps open at any given time than the XP interface...

Slow Slow 7

[Murdoch5]

I agree 100% with this article, I have 4 GB of Ram in my notebook and when I'm in WIndows 7 my memory consumption is somewhere around 3.5 GB, compared to Linux 2.6.32-r6 (Gentoo) which sits around the 512 MB mark. He's my question, what on earth does Windows need that much memory for? The OS should be taking up the least amount of memory possible to allow a user to run application that actually need the memory.

Re:Slow Slow 7

[Anonymous Coward]

You're a retard.

Congrats.

Re:Slow Slow 7

[1s44c]

I agree 100% with this article, I have 4 GB of Ram in my notebook and when I'm in WIndows 7 my memory consumption is somewhere around 3.5 GB, compared to Linux 2.6.32-r6 (Gentoo) which sits around the 512 MB mark. He's my question, what on earth does Windows need that much memory for? The OS should be taking up the least amount of memory possible to allow a user to run application that actually need the memory.

No. The OS should cache disk will all free memory because memory is fast and disk is slow.

Re:Slow Slow 7

[snemarch]

SuperFetch. Instead of looking blindly at "free physical memory", take a look at what SuperFetch uses and deduct that from the score, since it's essentially free ram - it's nondirty (aka clean) disk cache, and can thus be discarded right away if the system needs ram to service a memory request. And the up side is that SuperFetch speeds up application loading bigtime. I'd be pretty surprised if there aren't clones of it for linux already.

Re:Slow Slow 7

[Murdoch5]

oh probably lol, I'm not saying there aren't ways around the memory usage I just want to point out the fact that the memory is being taken up.

Re:Slow Slow 7

[Grem135]

of course it is being used....... and in the best way best way possible. I am one of those that has waaaaay to many windows open on machines for months at a time and even my 2 gig lappy that I remote to 3 other PCs, manage my WHS and use daily for everything but games and torrents runs flawless. BTW I finally rebooted it today for updates..... after 4 months Windows7 64 Turion64 x2 TL-64 2.2ghz 2gig ddr2 8400m GS 512meg video 15inch display @ 1280x800 250gig hdd I also have a AMD 4200+ running Vista 64 2 gig ram intel 6600 quad running Win7 64 4 gig AMD phenomx4 945 Win7 8gig ram P4 1.6 XP pro 768meg ram

Oh come on

[megla]

First we had submitters who didn't read the stories they were posting. Then we had editors who didn't read the stories they were approving. Now we have companies who don't read the articles they put out. Seriously, it's called a file cache. That's how it's supposed to work. Nice job, idiots.

Re:Oh come on

[DJoffe]

Did you read the article? It's not only memory, but also I/O and CPU:

"Both of those measurements are also higher for Windows 7 systems than for XP machines. While 85% of the former are running at peak I/O loads, only 36% of the latter do; the numbers for CPU workload are closer, as 44% of Windows 7 computers are running a computational backlog that delays processing tasks, compared to 36% of the XP systems."

A file cache alone isn't supposed to slow down your system --- if it is, "you're doing it wrong". Merely using more RAM may not slow down your computer but the article states explicitly that they are also measuring aspects that do impact on performance. Clearly there are more factors involved than just the file cache.

Re:Oh come on

[Anonymous Coward]

*ahem*

"resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks"

Re:Oh come on

[ShadowRangerRIT]

Vista and later OSes emphasize the I/O priority concept much more heavily than XP and earlier. File system indexing and defragmentation are done with ultra low priority; if *anything* else wants to read or write, they back off. Similarly, any service that starts as "Automatic (Delayed Start)" is opting in for low priority I/O. Unless their metrics say something about the slowdown of the active user processes, the fact that the file system indexer is backlogged is irrelevant.

Re:Oh come on

[Yaos]

The company that says this is happening also sells software that says it is happening? I believe them, in fact I'll buy 10,000 licenses right now.

Re:Oh come on

[Daniel Phillips]

Seriously, it's called a file cache. That's how it's supposed to work. Nice job, idiots.

The system is not supposed to slow down when the cache becomes full. Under load, Linux normally runs with all available memory used for cache and takes a negligible performance hit in that state.

New Windows OS != performance

[mnmlst]

First off, kudos to ComputerWorld for this shocking newsflash "New Windows Operating System is Bloated and Disappoints Users". Is it 1995 again when I foolishly believed Microsoft and loaded Windows95 on my happy Windows 3.1 computer only to discover the 4MB minimum RAM requirement left my computer a useless lump of plastic with an endlessly spinning hard drive? Four more MB of memory for $130 from a shady computer dealer finally slowed the paging down. I have seen this cycle repeated 6 more times since then. Go ahead and set up the fill-in-the-blank story for Windows 7.1, 8, and so on.

Here is how to get a valid test together:

1. Figure out the testing objective. Sounds like this guy should build a Windows 7 and a Windows XP box with identical hardware side-by-side.

2. Install the same applications on both machines and run the same workloads on them.

3. Measure the performance using the only benchmark users care about, waiting times for things to happen. One thing that was unclear from this article (which I actually read, must be new here) was the level of memory paging that was going on and especially the feedback from the users. The numbers he talked about are pretty much of no interest to end users, just guys in the I.T. shop.

4. Call ComputerWorld with the results, but only if they make Windows 7 look terrible...

Re:New Windows OS != performance

[shutdown -p now]

Measure the performance using the only benchmark users care about, waiting times for things to happen.

This was already done. You can assess the results indirectly by looking at Win7 sales figures (not preinstalled on PCs, but boxes).

Windows7 memory leak issues for ram even on 8GB

[Anonymous Coward]

Windows7 has wrost memory leak issues. Running scandisk can cause memory leak. Copy files on my socket 775 intel creates memory leaks. Not sure if its NCQ causing the leaks. Microsoft updates causing memory leaks??

When memory leaks starts it loses ram space hundreds of megs at a time until its done to 0mb left. Then the program start not to work with a white screen stuck at loading. Then can't open anything up saying out of memory. The whole time the computer turns from a 4ghz super fast machine to a many thousands of dollar paper wieght 286 machine.

 

Re:Windows7 memory leak issues for ram even on 8GB

[Pojut]

If you are running a PC at 4 GHz, you are either running a Pentium 4 (in which case your technical knowledge is questionable at best) or are running a very overclocked system (in which case you would be smart enough to use something other than Windows 7 if it was causing you that much grief).

Considering your "4 GHz" claim combined with "many thousands of dollar" for your PC, I'm going to go ahead and call BS on you.

Re:Windows7 memory leak issues for ram even on 8GB

[Anonymous Coward]

I have many system and I can atest to my original post. Its a Intel core 2 duo at 4ghz. I said what I meant memory leak. Its not a issue a non-technical person like your self understands. I'm a IT profession for a living and a regular slash reader for more years than you been born I'll leave it at that. Reporting memory leaks isn't rocket science du_bass.

Re:Windows7 memory leak issues for ram even on 8GB

[Pojut]

Quoted from a post I made earlier in this article:

People are either being unbelievably stupid, have only 1 gig of RAM installed, or this is FUD. Example: I'm currently running Windows 7 64 bit. On my secondary monitor, I have a bunch of system monitoring widgets... hard drive space, CPU load and temp, video card load and temp, memory usage, etc. Just last night I was playing Bioshock 2, all settings at max. Even with those widgets running, with Aqua Teen Hunger Force playing in MPC on the secondary monitor in a window, and Bioshock 2 running full bore, I was still using only 52-58% of my available 4 gigs of ram. I call BS on this article.

Here are my system specs, to back up my claims. As you can see, nothing special (copy and pasted from my [H]ard|Forum sig):

Display: Asus VH236H | Dell 2005FPW
Foundation: Cooler Master Storm Scout | OCZ ModXStream Pro 700w
System: Gigabyte GA-MA785GM | AMD Athlon 64 X2 5400+ | Corsair XMS2 4GB DDR2 800 | ATI 4850
Internal Storage: Diamondmax 21 system | WD15EADS archives
External Storage: 1.25TB in a KINGWIN DK-32U-S | WDMER1600TN
Input: Kensington 64325 Expert Mouse | Saitek Eclipse II | M-Audio Axiom 25
Audio: Logitech Z4 2.1 | Audio Technica ATH-AD700

For the record, I do mail merge programming for a living and have a collection of old systems ranging from a (fully functional) TRS-80 all the way up to my current system.

Re:Windows7 memory leak issues for ram even on 8GB

[Fantastic Lad]

Considering your "4 GHz" claim combined with "many thousands of dollar" for your PC, I'm going to go ahead and call BS on you.

Yeah, or he's a regular consumer who spent a lot of money on his computer and simply believed the twenty-two year old in the tie who added the clock speed of each core of the multi-core computer together while promoting the sale. After all, when you look at MS's "System Properties" it does that math right there on the screen for you and you feel all special.

-And apparently, if you use Windows 7, rather annoyed as well.

-FL

Re:Windows7 memory leak issues for ram even on 8GB

[Anonymous Coward]

You're a terrible troll.

Re:Windows7 memory leak issues for ram even on 8GB

[zrq]

System: Gigabyte GA-MA785GM | AMD Athlon 64 X2 5400+ | Corsair XMS2 4GB DDR2 800 | ATI 4850

According to wikipedia the CPU clock speed for an Athlon 64 X2 5400+ is 2800 MHz not 4GHz.

Re:Windows7 memory leak issues for ram even on 8GB

[Pojut]

I never said my system was 4 GHz...you are confusing me with the AC OP on this thread.

Re:Windows7 memory leak issues for ram even on 8GB

[Anonymous Coward]

You've never met a PC gamer have you?

Re:Windows7 memory leak issues for ram even on 8GB

[Pojut]

I (and most of the nerds I know) are PC gamers...and none of them would use an OS that ran unstable on their hardware.

Re:Windows7 memory leak issues for ram even on 8GB

[zrq]

My apologies.

Re:Windows7 memory leak issues for ram even on 8GB

[Pojut]

No prob, it happens :-)

Parent is +1 informative

[Anonymous Coward]

You cannot study virtual memory performance without considering how many page faults occur.

It is perfectly reasonable to use RAM as a filesystem cache, which is why Linux has adopted this approach. The effect is that almost all of the physical RAM is always in use. The cost is that pages are more likely to be wrongly swapped out - however, in typical cases, this increased cost is tiny in relation to the huge reduction in the number of disk accesses.

Re:Parent is +1 informative

[Anonymous Coward]

Unfortunately, just measuring "page faults" isn't a very reliable indicator either. Reading in an executable from disk (because you're, say, /starting a program/) incurs page faults to the .exe and any .dlls that aren't already cached. Reading a data file from disk? Page faults.

If you want to measure activity on the page file, afaik, the only way to do so is to isolate the page file on its own physical hard drive all by itself, then point your performance counters at it.

Re:Parent is +1 informative

[Anonymous Coward]

Agreed, but also remember that Windows' page fault count includes soft faults, that is faults serviced out of cache or out of another process's working set.

Windows memory management is quite complicated. You can't understand what's going on just looking at one or two headline figures.

Re:Parent is +1 informative

[rickb928]

"It is perfectly reasonable to use RAM as a filesystem cache"

Especially if you can tolerate losing data in a power failure, or just plain old system halt.

And that never happens to Linux machines, much less Windows 7 machines.

Smartcache used to be manageable, sort of. Sounds like it's pretty aggressive now.

Write caching is the next step, where it really, really hurts.

I asked my Linux buddy about this, and yep, he sees 80-90% of his 8GB of RAM in use on his Windows 7 machine. With a browser instance and Freecell open, full-on Aero, and nothing else except corporate-mandated Symantec 'protection'. He's interested in this all fo a sudden. 5-6GB file caching for this? He's shocked, shocked!

Somehow, using all available RAM for file caching seems to me to be leading to a potential RAM-thrash as you load up some useful applications. So please, educate me, and explain how this is not a problem. I can hammer my XP machine here with 3 big virtual machines and Lotus Notes/McAfee hogging 500MB in a 4GB machine, and it still has some 500-600MB RAM left over. No big performance problems. I have more trouble with performance when I load /.

Re:Parent is +1 informative

[Jeremy Erwin]

Here's how MacOSX measures memory (this is from top, btw)

PhysMem: 844M wired, 1035M active, 378M inactive, 2258M used, 1838M free.
VM: 138G vsize, 1041M framework vsize, 50715(0) pageins, 0(0) pageouts.

If you have more than a trivial number of pageouts, you need more memory.

Re:Parent is +1 informative

[drsmithy]

Somehow, using all available RAM for file caching seems to me to be leading to a potential RAM-thrash as you load up some useful applications. So please, educate me, and explain how this is not a problem.

As you load "useful applications", cache memory is freed for them. Surely, this is obvious ?

The last mainstream OS to have a static cache was MacOS Classic. Before that, it was OS/2. Windows has had a dynamic disk cache since 1995 ('92 if you want to include Windows NT).

Haven't seen that yet

[nurbles]

I've only been running Windows 7 x64 at home for about a month now and I've rarely (if ever) seen it use much more than 50% of available RAM (I have 8GB). Then again, I don't run any software from Microsoft (except Windows itself) or Apple or any other mega corp that thinks they own your machine after you install their word processor or music player or whatever. With just a few games, Firefox, Thunderbird, OpenOffice and vlc, my Windows 7 box is much better than it was with Vista x64 and at least comparable to when it was XP (but it boots MUCH faster). Even when I had the BOINC client installed and running climate prediction and seti@home (both regular and nVidia GPU versions), the memory usage was quite reasonable.

The article doesn't appear to differentiate between the 32 and 64 bit versions or what kind of app/usage mix the machines have, which must figure in to this sort of measurement. It also isn't at all clear whether the Win7 systems are running programs built for Win7 (as the XP systems almost certainly are) or XP builds that are "runnable" in Win7, though it is quite possible that could have a significant impact.

I've been developing for Windows for 20 years and in my experience, it has almost always been bad applications and/or 3rd party drivers that cause Windows to get such a bad rap. As a counter to the typical Windows bashing, I'd like to point out that I have one customer that has been running our software on Windows NT4 machines running 24x7 for over 10 years (most of which went for 5 years or more without rebooting at all) and the only unintentional down time came from hardware failures. Another customer is using Windows 2000 (and now XP) the same way.

What a bunch of drivel.

[Anonymous Coward]

I find it interesting the author mentions a "new" metric as part of their quantifying this problem. Huh? Has anyone tested this metric? Or that current generation hardware is being maxed out. I obviously am in the minority since my computer is in the 14% that doesn't have this problem and so is my parents laptop. Funny how just in my family we are in the minority. Been awhile since I've done statistics, but I think that works out to 1.9% chance of having two computers not having the problem. This article sounds very suspect.

In XP? Definitely YES

[Joce640k]

FIle cache will definitely swap out your applications in XP.

In previous Microsoft OSs you could set maximum file cache via the Windows .ini file (and it was like a breath of fresh air for Windows performance - suddenly you could do other things while burning CDs, etc.

On XP they took out that feature and it's performed like a dog ever since because of it. A whole new generation of CD burners had to be developed with "SafeBurn" technology, etc.

I'm sure the stupidity has continued in Vista/7 but I don't have it installed on any of my work machines so I can't be 100% sure. If it has page file right after booting up then you can be sure it does.

Re:In XP? Definitely YES

[Anonymous Coward]

BURNproof was not a result of a change in the win.ini file. Ever since the first CD writers came on the market the common PC had huge problems delivering a sustained data stream because of fundamental weaknesses in the hardware architecture, the PATA interface being the number one problem. Windows "multitasking" issues did not begin with XP.

Re:In XP? Definitely YES

[DJoffe]

FIle cache will definitely swap out your applications in XP.

I don't know about Windows 7, but I can also definitely confirm that older versions of Windows do this, including XP, and not even remotely "intelligently". I used to often have to copy large amounts of files over the network via SMB, and you can do a very simple test to confirm this: (a) Load up some applications (b) Copy (with that machine as SMB server and the receiver as client) a Gb or two of files over the network (c) bam, your applications are all swapped to disk and visibly, painfully crawling along --- they will pretty much be visibly more sluggish until you restart. Windows memory management is awful, or at least it used to be --- its quite possibly they've improved it in Windows 7. Windows used to have other major deficiencies with memory management, such as over-aggressively starting to swap to disk when only around half your actual RAM was used ... again, it's possible this is improved in Win7, I haven't done work requiring testing those aspects in some years.

Re:

[Anonymous Coward]

Hey samzenpus, kdawson has been using your username!

They Report Disk I/O Backlog Percentages

[eldavojohn]

Barth acknowledged that XPnet's data couldn't determine whether the memory usage was by the operating system itself, or an increased number of applications. So yeah, it doesn't seem like the author really knows what's going on...

While that's true, one would probably make the assumption that it is normalized in XP vs Windows 7 since they have no way of tracking it. What I mean is that you would assume the Windows 7 user runs the same number of programs as the XP user.

I actually followed the blog link in the story and while they can't pin it on application or OS, they can say that the disk I/O is backlogged on 36% of XP machines sampled, 83% of Vista machines sampled and 85% of Windows 7 machines sampled.

While they don't know anything about the applications being run, this backlog is probably how they determined that processes were being forced to resort to virtual memory running on the disk. Is there a better explanation for those numbers?

Re:They Report Disk I/O Backlog Percentages

[mystikkman]

Windows 7/Vista do search indexing, defragmentation in the background at low disk priority.

Re:They Report Disk I/O Backlog Percentages

[kenp2002]

Windows 7/Vista do search indexing, defragmentation in the background at low disk priority.

Not only that but also obeys NCQ\TCQ reordering, and auditing is enabled on Vista and W7 along with Last Accessed timestamping (so any AV, defragging in the background will trigger additional IO ops to update said timestamps on literally any file that gets touched.)

Disable last access time stamps, disable NCQ\TCQ support (unless you have a real workstation or server) and disable auditing and watch that backlog vanish.

Re:They Report Disk I/O Backlog Percentages

[compro01]

The I/O backlog on vista/7 isn't very comparable to XP's. IIRC, vista/7 natively supports native command queuing, which queues and sorts disk accesses to make them as sequential as possible. This grows the apparent backlog, but increases throughput and average latency when doing lots of small random seeks, which is a very good thing IMO given how ludicrously slow hard drives are at this.

As usual, this is slightly non-optimal when the assumption (lots of random seeks) is not met and it can be beneficial to disable NCQ under workloads that use lots of huge sequential operations.

Re:They Report Disk I/O Backlog Percentages

[Nefarious Wheel]

Size of the disk IO queue? Even small integer values are bad.

Idiotic article

[Anonymous Coward]

I'm an engineer, and this article is total fluff buillshit. You want to post something like this, I want METRICS! I want test groups, system configs, things like page faults, pages/sec, RAM total, installed apps, running apps, et al.

Wlecome to the 5 second soundbite.

Morons.

From TFA:

[falconcy]

Users who want to compare their computers to the current WCPI numbers can do so by registering with XPnet and then installing the DMS Clarity Tracker Agent from Devil Mountain's site.
Nuff said

This article is QFT. Qouated for truth.

[Anonymous Coward]

Lots of people are non technical on this slash form which use to be great now I can't even found my posts which means its a very bad site.

First people are complain about a issue. The people posting microsoft technical advertisement are not experienced.
When there is a real issue this article complains about it. Which in term I can atest to the issue. I also encounter various situation were memory leak happens in windows 7 been using it since it came out. Its is even been reported on microsoft problem site. If people want to look it up.

People who are denying it and posting counter arguement doesn't seem to understand there is a issue. Yes I use it and the memory leak doesn't happen all the time, but does happen and happens with regular use.

Memory cap

[1s44c]

The 32 bit memory cap is per process, not per system. PAE has been on all 32bit intel and amd processors for a very long time.

Re:Memory cap

[Nzimmer911]

Sorry dude, from Microsoft: Physical Memory Limits: Windows 7 The following table specifies the limits on physical memory for Windows 7.
Windows 7 Ultimate 4 GB 192 GB
Windows 7 Enterprise 4 GB 192 GB
Windows 7 Professional 4 GB 192 GB
Windows 7 Home Premium 4 GB 16 GB
Windows 7 Home Basic 4 GB 8 GB
Windows 7 Starter 2 GB 2 GB

Re:Memory cap

[andrewagill]

How many motherboards support that?

It doesn't matter how much RAM the chip can support, if your motherboard only supports 1GB of physical RAM, and the rest has to be virtual memory, which can result ``result in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks,'' according to TFA.

Re:Memory cap

[blueg3]

However, PAE drivers have to use 64-bit physical addresses, or they won't work properly. In testing, Microsoft found that too many drivers crashed on PAE systems with more than 4 GB of memory, so they limited usable physical address space to 4 GB on PAE systems.

Re:Memory cap

[andrewagill]

Version | Limit in 32-bit Windows | Limit in 64-bit Windows:
Windows 7 Starter | 2 GB | 2 GB

Seriously? Microsoft, you brought this on yourself.

Re:Memory cap

[ShadowRangerRIT]

Starter edition is an intentionally crippled OS used to lower their prices in developing countries without undercutting sales of the full OS to those that can afford it. You may disagree with it on personal principle, but it's a sound marketing strategy.

Re:Memory cap

[andrewagill]

OK, now it sort of makes sense, but I still think they should cap it at at least 4GB for 64 bit--otherwise, what's the point of a 64-bit chip?

And yes, I realize the irony of writing this on a 64-bit system that only has 2GB of RAM, but the motherboard claimed that it supported 4GB.

Re:Memory cap

[1s44c]

I didn't believe it at first but it turns out you are totally right. Even my home ubuntu system has 8 gig of ram.

It turns out that every time I think 'even Microsoft can't be that bad.' I get proved wrong.

Re:Memory cap

[1s44c]

However, PAE drivers have to use 64-bit physical addresses, or they won't work properly. In testing, Microsoft found that too many drivers crashed on PAE systems with more than 4 GB of memory, so they limited usable physical address space to 4 GB on PAE systems.

Drivers on 32 bit Linux with 8 gig of ram work fine.

Re:Memory cap

[blueg3]

There are many, many few third-party, binary-only drivers for 32-bit Linux, and I'd be willing to bet that a number of them don't work properly on PAE systems.

It's not that you can't have 32-bit PAE systems with more than 4 GB of memory, it's that precompiled third-party Windows drivers crash on such systems.

For that matter, a process can only use 2 or 3 GB of memory, since the kernel shares address space with the process. And the system as a whole can't have a full 4 GB of memory -- it's restricted to 4 GB of address space, some of which is consumed by device memory.

Note that some versions of Windows 2000, Windows Server 2003, and Windows Server 2008 lift this fake limit, allowing you to use more than 4 GB physical memory on 32-bit systems (with PAE).

This is all covered in Windows System Internals. You can also read a bit about the driver issues here.

Re:Memory cap

[Mad Merlin]

How many motherboards support (PAE)?

All of them since the late 90s?

Re:Memory cap

[andrewagill]

All motherboards since the late 90s support more than 4GB of RAM?

Re:Memory cap

[1s44c]

There are many, many few third-party, binary-only drivers for 32-bit Linux, and I'd be willing to bet that a number of them don't work properly on PAE systems.

It's not that you can't have 32-bit PAE systems with more than 4 GB of memory, it's that precompiled third-party Windows drivers crash on such systems.

For that matter, a process can only use 2 or 3 GB of memory, since the kernel shares address space with the process. And the system as a whole can't have a full 4 GB of memory -- it's restricted to 4 GB of address space, some of which is consumed by device memory.

Note that some versions of Windows 2000, Windows Server 2003, and Windows Server 2008 lift this fake limit, allowing you to use more than 4 GB physical memory on 32-bit systems (with PAE).

This is all covered in Windows System Internals. You can also read a bit about the driver issues here.

The system as a whole can't have more than 4GB of memory with PAE? What do you think PAE is for then? It's obvious that pointers are still 32 bit, you don't have to restate it.

The 32 bit/4 Gig system ram limit was fixed in hardware and in serious OS's a long time ago. Windows isn't keeping up with the basics.

Re:Memory cap

[blueg3]

You don't seem to be listening. I'm well aware of what PAE is for. As you seem to admit -- "Windows isn't keeping up with the basics." -- I'm simply pointing out that there is, in fact, a 4 GB memory limit in (most) versions of 32-bit Windows, PAE or no, and it exists for technical reasons. The technical reason is not "it's not possible to address more than 4 GB with PAE", but it's still a technical reason.

The problem is that the pointers *aren't* all 32-bit. Virtual address pointers are, so 32-bit processes don't have to care if it's PAE or not. Physical address pointers are longer than 32 bits. Drivers use physical address pointers. Drivers care whether you're using PAE. It's an easy problem to address if you can just recompile the driver, but this isn't an option for Microsoft. As a result, they limited physical memory on 32-bit systems to only 4 GB, even on PAE systems, so that even physical address pointers will fit in 32 bits. It's straightforward to understand and documented.

This is why you shouldn't run 32-bit Windows, even though it supports PAE.

Re:Memory cap

[Mad Merlin]

No, PAE.

Re:Memory cap

[petermgreen]

MS disabled use of address space beyond 4GB on 32-bit desktop windows with XP SP2, they claimed this was because of bad drivers. There may be some truth in this (XP SP2 was also the version where they enabled PAE by default for other reasons (data execute protection IIRC) or it may be an attempt to push users to 64-bit or server editions.

Currently the only supported 32-bit desktop edition of windows that hasn't been crippled in this way is windows 2000 and that won't be supported for much longer. IIRC there is still a memory cap at 4GB of memory (rather than 4GB of address space) so using a desktop edition of windows without the PAE crippling doesn't buy you that much in most cases.

There is a hack to enable all your memory in vista and win7 but it involves hex-editing the kernel and then signing it with your own signature and booting with test-signing enabled.

Using a sufficiently high server edition also gets round the crippling but that's an expensive option.

Re:Memory cap

[petermgreen]

Linux has a very different driver ecosystem to windows.

On Windows most drivers are developed by the hardware vendor and distributed as binaries to the buyers of the hardware and maybe to windows update. This means there is little quality control (some drivers get submitted to WHQL but many don't and even when they are I don't think MS gets the source) and little to no ability for anybody else (including MS) to update/fix the driver if the hardware vendor can't be bothered to. Writing new drivers from scratch is a possibility but few people are prepared to write Windows drivers if they aren't being paid to do so ;).

On Linux most drivers (regardless of whether they are sourced from a hardware vendor or community written) are maintained and distributed as part of the Linux kernel source tree. This means that bugs can be fixed and ports can be made by anyone in the community that cares enough to do so. On the downside the lack of a good method for distributing Linux drivers seperately and loading them into the installer for any distro often makes using Linux on very new hardware painful.

Re:Memory cap

[blueg3]

Yes -- since the 4 GB restriction on PAE is entirely in-kernel, it can be bypassed (since full PAE capabilities are actually supported, for the datacenter versions). You're then on your own for driver safety, but one would hope that anyone with 64-bit drivers would be releasing PAE-safe drivers.

Given how badly-written Windows drivers are, I can believe their explanation.

Re:Memory cap

[andrewagill]

OK. To recap, if you don't have more than 4GB of RAM, PAE is not going to give you more than 4GB of RAM.

TFA is saying that users are maxing out their physical RAM and have to use swap. Quoting:

On average, 86% of Windows 7 machines in the XPnet pool are regularly consuming 90%-95% of their available RAM, resulting in slow-downs as the systems were forced to increasingly turn to disk-based virtual memory to handle tasks.

My point was that even if you have a processor that supports PAE and even if your motherboard supports PAE, you can't use it unless you have more than 4GB of memory, and even then, if you don't have 4GB of RAM, PAE will not affect the problem that TFA is reporting on.

Re:Memory cap

[Mad Merlin]

OK. To recap, if you don't have more than 4GB of RAM, PAE is not going to give you more than 4GB of RAM.

I agree completely, but I never said anything about 4G of RAM, or any quantity of RAM at all for that matter, I was only talking about PAE.

My point was that even if you have a processor that supports PAE and even if your motherboard supports PAE, you can't use it unless you have more than 4GB of memory, and even then, if you don't have 4GB of RAM, PAE will not affect the problem that TFA is reporting on.

Not for addressing extra memory no, although these days you can use PAE to emulate NX.

As for TFA, it's downright silly and not even worth discussing. The author has obviously never heard of caching, or is willfully ignorant in the name of marketing. Someone should make a Windows version of this site.

Re:Memory cap

[1s44c]

Linux has a very different driver ecosystem to windows.

Linux drivers work on 32 bit PAE, windows drivers don't.

32bit windows is a joke.

Re:Memory cap

[1s44c]

Yes I agree there is a technical reason. The reason is that windows drivers are broken. The technical answer is to fix the drivers.

For the most part Microsoft knows who these driver writers are. They should be contacting them and working something out instead of pretending that it's an impossible task. They have had many years to fix this but choose to tell their customers it's impossible and they have to upgrade to 64bit.

8 Gig of ram and a motherboard that supports it is now within the budget of most employed people. Maybe some of those people have to use 32 bit windows. If windows can't support core system hardware like memory there really isn't much hope for it.

Re:Memory cap

[toddestan]

You'll also find 7 Starter on some of the netbooks sold in the US.

Why in the name of all that's holy...

[ricky-road-flats]

...is anyone in the know still using swap?

I've not used swap on any desktop, laptop or server for several years now, and not had a problem. Domain controllers, Exchange servers, SMTP relays, file servers, web servers, database servers. Windows, Linux. 32 bit and 64 bit. All fine, and some with uptimes over a year. I agree there will be outliers who need it, but most people and most situation's don't.

Provide enough RAM for what you need and switch swap off!

On a desktop/laptop, towards the end of a busy day with lots of apps open, see how much you're using and most importantly what the peak utilisation was. If it was more than your RAM, you need more RAM. If it's comfortably below the RAM you have, you can switch off swap and get a free significant performance boost.

The performance gains alone should make it worthwhile, let alone your reduced hard drive wear and tear, and power savings (hard drives can go to sleep much more frequently and for longer) on top of that...

Oh, and as someone else has correctly pointed out, RAM that's in use for caching is not being wasted, the RAM that's sitting there not being used that's being wasted.

Re:Why in the name of all that's holy...

[andrewagill]

I once wanted to do exactly this, but... I make the assumption that the kernel programmers take it for granted that you have swap, so I provide it. The next PC I make may have a DRAM-based hard drive dedicated to swap.

Re:Why in the name of all that's holy...

[slim]

Here's a good writeup from a developer who definitely expects you to be using swap:
http://varnish-cache.org/wiki/ArchitectNotes

In brief, if you want Varnish to keep a 8GB HTTP cache, then it's going to malloc() 8GB. It expects the kernel to manage swapping whatever parts of that are unused.
He points out that in an app such as Squid, which tries to maintain its own disk cache alongside its RAM cache, the two swap algorithms fight each other in a very counter-productive way.

I think all developers ought to be free to take that same approach. Why should an application programmer write potentially substandard memory management routines when the OS is supposed to be good at it. I agree there will be outliers where the application has special domain knowledge that it can use to manage memory more effectively than the kernel. I suspect there are *very* few such cases.

Re:Why in the name of all that's holy...

[ricky-road-flats]

Except that assumption has proved not to be true in my wide experience... Windows (XP, Server 2000, 2003, 2008, Vista, 7) is quite happy without swap, and it is an allowed setting, Linux (from 2.4 onwards) you can either simply not mount any swap, or if you want to get really anal about it (I never have needed to) compile the kernel without swap support.

There was one app which REQUIRED swap to be present, and on those couple of machines a fixed-size 16 MB (the minimum allowed above zero in Windows) has solved it. That was an old version of AutoCAD on Windows XP.

Try it, you'll see. You certainly don't need to spend significant money to have something you almost certainly don't need, spend the money on more RAM and a fast CPU instead!

Re:Why in the name of all that's holy...

[ricky-road-flats]

A fair point, and when doing any new server software I always do testing before going live, naturally. As I already pointed out there will be exceptions to the rule, but it's not a theoretical discussion, I've done it.

I was head systems manager in a setup of over 280 servers and 3,500 clients for a few years, and from 2003 until I left in 2009, only one of those 3,800 machines (a legacy AIX processing box) needed 'normal' amounts of swap. There was one version of AutoCAD which whinged without swap, so the 2-3 machines that ran on got 16 MB (the minimum non-zero amount Windows will allow), and with that they were fine.

Re:Why in the name of all that's holy...

[andrewagill]

I presently have swap set to 512MB, the minimum recommended swap size. If an app needs swap, it's there, and they can have it. Otherwise, I have 2GB of RAM and I'm using 62MB of that swap.

Re:Why in the name of all that's holy...

[Blakey Rat]

You can't memory-map large files if you have swap turned off. If you don't run any applications that do this, well... fine. But if you do, turning off swap is a huge performance hit.

Slow migration to 64 bit?

[Anonymous Coward]

Could this be a symptom of slow migration to 64-bit Windows? 32-bit Windows 7 can use about 3.5GB RAM at max. Time has come for mass migration to 64-bit...

Linux standard practice

[DrYak]

BeOS and Haiku did/do this, but I don't think any other OS has implemented total RAM usage to such a degree.

It's also a standard practice in Linux too (and *BSD, and their Mac OS X derivative and probably any other modern Linux).
That's why, when you use the "free" command on these OSes you get two lists of number.
Total memory usage, and how much of that memory is actually used by processes and who much is cache.

- Now this memory should be dynamically freed when the process require more resource.
- In addition to that, un-/seldom- used process memory can be swapped out and more memory made free for caching. (Modern OSes can do it to some extend with various settings of aggressiveness : there's no point keeping into RAM a piece of data that will never be accessed, better swap it out and use the RAM estate for caching to accelerate the whole system).

BUT
If, as the summary states, Windows starts thrashing the swap, then it means it's really low on memory.

I personally suspect it's the "Vista-ready" debacle all over :
TFA mentions that Win7 machine have 3GB average. Only. Whereas, from my personal experience, 4GB is the minimum needed for comfort.
Thus probably a lot of machine are shipped by their manufacturer with the bar minimum of memory to be able to boot into Win7.
Just like some Vista-ready machine only had the strict minimum to install and boot into Vista with all features disabled.
Or the first WinXP machine which shipped with only 128MB (sometimes 256MB) of RAM.

Of course, it doesn't help that Windows OSes are such resource hogs to begin with. Just install the latest Xubuntu or other light-weight oriented distro and the machines breathes again. And that's a current-day general purpose distro. Now, with things like Damn Small Linux, you can bring back from the grave even more resource starved machines.

Re:Linux standard practice

[amliebsch]

You don't need 4GB of RAM for Win7. I run Win7 - with Aero! - on a Dell Mini with an Atom processor, 1GB of RAM, and a 16GB SSD. It doesn't have the horsepower to play youtube videos smoothly, but web surfing, email, MS Office all run just fine. Even Slashdot!

That's not been my experience...

[_Shad0w_]

My entirely non-scientific appraisal of Windows 7 thus far is showing it using less memory than Vista was. Right now it's using 32% of the 8GB of physical RAM. Vista would generally be running at around 50%, doing the same thing, which right now isn't much beyond IRC, FF with three tabs, Thunderbird and things like NCSoft's Launcher and Steam sat in the notification area.

I'd expect some difference due to differences in running software, but for the most part I have the same things running on Windows 7 as I did when I had Vista on the system.

W7 memory leak bug - recently hotfixed

[Anonymous Coward]

Sorry for the AC post but I was having serious memory issues with win7 up until recently. My 2GB RAM laptop booted up using 700-800MB but after leaving it on for 24 hours or more it crept up to 90-98% memory utilization. Taking my laptop down to a crawl. Turned out there was a bug in Win7 where the Windows Filtering Platform (WFP) callout driver could cause a memory leak. A hot fixed was released: http://support.microsoft.com/kb/979223 and it cured my problem. I wouldn't be suprised if that bugg was resulting in the huge memory consumption seen from the study

Windows systems are sold without enough memory.

[clone53421]

News at 11.

You want to upgrade your memory? You really should. It’ll cost extra, but your computer will perform much better if you do. Yes, we could sell it with enough memory to begin with, but then we wouldn’t be able to advertise a low low price and then charge you extra for the memory upgrade...

The whole “pay extra” memory upgrade makes people think it’s an optional thing, and to save some money they skip it.

Seriously. I’ve “fixed” XP systems that were purchased with 256 MB of RAM. They had faster processors than my computer had, but of course they’re going to be slow as hell... what did you expect? The “solution”? Go to Micro Center and buy a fucking stick of RAM. A gig will do fine. And no, I’m not paying $40 to have you open the case and pop it in.

Not a "Windows 7" thing...

[adosch]

At least on the 64-bit comparison, anything (e.g. Windows 7) that has a "minimum" memory requirement of 2GB in terms of an OS, what did you expect? Not that having 2-4GB in any sort of laptop or desktop system these days isn't uncommon, but I don't think I exceed 2GB in anything I currently own, laptop or PC wise. Quite honestly, I don't run Windows 7, and I've been a Linux convert for some time now on the desktop, but I see the very same thing with over-bearing GUI'fied eye-candy Window Managers on Linux, too. I'm constantly blowing away my pagecache on under Gnome and making kernel parameter adjustments to see what little improvements I can make.

Unless I'm running LXDE, icewm, fluxbox, ect. and keeping GUI's down to my browser and the handful of xterm's I've got open, I've found it hard to keep a balance of memory usage as well. The days of lightweight GUI anything are over.

And no, this isn't a lure for a Linux fan boi X-Windows + Window-Manager-of-the-day debate.

The solution

[rcharbon]

Maybe the Win7 users should stop running apps that are pure overhead? You know, like the XPnet client?

Windows 7 does report used memory not cached in th

[joaeri]

Bring up the task manager. On the "Performance" tab select "Resource Monitor..." go to the "Memory" tab. There you will see what the memory is used for and you can see that "standby" memory is not reported as used if you check the task bar. A tip is also to just hover the mouse over the colored bar to see what that memory is used for in more detail.

Had a similar problem with XP

[ymgve]

I haven't tested Windows 7 yet, but I was always annoyed that XP seemed quite fond of prioritizing disk cache over background applications. Every time after I had been playing a game, my web browser/email client would be slow as molasses because it was completely swapped out and it would take minutes before it went back to normal speed. So I did the one thing that a lot of people think is some kind of blasphemy.

I turned off swap.

And never looked back. I don't have to worry about disk trashing when changing between programs anymore, the whole system feels faster and I've never run out of memory, despite only having 3GB. Something was definitely flawed in the XP memory management.

stop the press!

[garry_g]

This just in: New Windows uses more RAM than previous version! Experts are surprised!

Is it just a dejavu I'm having, or was this the standard behavior for every single version of windows?

Windows 7 makes maximum use of unused memory

[FirstTimeCaller]

Yeah, I guess that headline doesn't sound as sensational...

Comment removed

[account_deleted]

Comment removed based on user account deletion

Windows and Linux... more alike now

[Anonymous Coward]

Windows 7/Server 2008 have finally caught up to Linux in that they use free memory for disk buffers and cache. When you run free -m, you see "mem" and then "-/+ buffers/cache". Add "free" values for "-/+ buffers/cache" to "mem", and you have the "real free memory." With Windows 7/Server 2008, go to Task Manager/Performance tab. Add "Cached" value to "Free" and you get the "real free memory"... it's just in use for cache since it's not needed for another application. This is the same way Linux works.

Windows ALWAYS swaps no matter what. This is still different from Linux.

I thought it was 87%

[RudySolis]

Studies have shown that accurate numbers aren't any more useful than ones you make up."


http://1.bp.blogspot.com/_oqH68z1KYWk/SCMhkz0851I/AAAAAAAAAGQ/7hQDeVNcikA/s1600-h/5652.strip.print.gif

So?

[Sj0]

Who cares if the PCs are maxing out their memory if the user experience is equal to XP in terms of perception of speed?

RAM goes where?

[tepples]

If I ever find it suffering due to lack of memory, I shall buy more.

And stick it where? A lot of PCs, especially low-end laptops and older desktops, don't have space for more RAM.

Re:RAM goes where?

[theaveng]

As technology advances you can usually find more RAM squeezed on a single "stick" of RAM.

For example my old Windows 98 laptop had only 32 megabytes plus 64 megabyte expansion when purchased new. At the time it was the max available, but now ~10 years later people have squeezed 512 megabytes per stick. So my ancient laptop now has enough room to run Windows XP (nice and fast), Ubuntu Linux, or even Windows 7 (slow but it works).

Re:RAM goes where?

[Lumpy]

Mid end newer ones dont as well. My Dell studio 17 that I just got has a MAX of 4 gig. it cant take any more than that.

And that is not enough for most of what I do.

Re:RAM goes where?

[Anonymous Coward]

then you're stupid for buying a machine inadequate for your needs.

On a laptop, 4 gig is probably double what a average user would ever need.

Re:RAM goes where?

[petermgreen]

As technology advances you can usually find more RAM squeezed on a single "stick" of RAM.
You do but you often find that motherboards are limited in not just the number of ram sticks but the size of them too. Sometimes you can go over the official maximum sometimes you can't.

Because of the fairly frequent changes in ram types I can't just borrow a modern stick from some other machine to try so I either have to stick to the manufacturers official limit or take a risk on ordering an expensive (modules at the large end of thier technology are expensive in my experiance) stick of ram that may or may not be supported.

And a motherboard upgrade can get expensive particulally if it involves replacing other parts and/or MS refuse to activate your OEM copy of windows on your new board.

Non-paged pool?

[Anonymous Coward]

There was some discussion about disk cache inflating memory utilization stats, but to me, swapping is the real test. When paging/swapping takes place, the speed of the computer is pretty much limited to the throughput of the disk, which is an order of magnitude slower than even the slowest memory access. As an added bonus, most Windows machine have only one physical hard drive, so the swapping is competing with system and user I/O that is probably the root cause of swapping in the first place.

A little bit of swapping is good for you. A typical machine runs quite a few services in background but are used less than 1% of the time. I don't print very much. If the print spooler is swapped out, no biggie. Even if it has to get swapped back in, the performance penalty is nothing compared to the warmup time for the printer.

I am not a Windows 7 expert by any means, but I know about VMS, which is where Microsoft got many if its ideas for Windows - especially memory management. VMS was ultra-stable -- perhaps MS should have borrowed a little MORE from that technology. DEC had numerous patents (having co-mingled software and hardware) and held on to some of their top OS people until the bitter end. Thus, Microsoft was unable to make Windows a complete ripoff of VMS. It wasn't for lack of trying.

As we search for viable explanations for apparent memory saturation (beyond disk caching), consider this great article on paged and non-paged pool

Junk characters

[tepples]

I had a great reply all typed up but the stupid filter thinks it used too many "junk" characters... not entirely sure what those junk characters are

On Slashdot, "junk" characters are characters that are more useful for ASCII art than for prose. The filter dates back to when spammers used to post ASCII art representations of gay pornography.

Re:Junk characters

[mujadaddy]

On Slashdot, "junk" characters are characters that are more useful for ASCII art than for prose. The filter dates back to when spammers used to post ASCII art representations of gay pornography.

Before the dark times...before the Empire...

Re:Junk characters

[lagfest]

So they are called Junk characters because they are used to portrait a naked mans penis? What a coincidence :)

In other words, 86% did not disable superfetch

[postmortem]

Which is left on by default.

And that "bad performance" simply meant that user opened app that was not already cached in RAM.

But Slashdot filter was pleased: if bad article about windows submitted, then publish it on front page

Re:In other words, 86% did not disable superfetch

[slim]

"Bad performance" wasn't even observed.

They somehow got their tool onto a bunch of PCs. They collected the results. They saw high memory use stats. They sucked in their breath and said "Ooh, if there's no spare RAM, performance will suffer". Then they started typing into their word processor.

Anti-Virus progs

[Hierophant7]

I noticed this problem myself months ago. After a few days of uptime, the RAM would fill up by itself (not by me opening more and more programs... I would come home from work or wake up in the morning and my RAM usage had gone up considerably) and the machine would become quite slow once the RAM usage hit 85-90%. And this PC has 6GB of RAM, so that's not a trivial amount of memory usage. After fiddling for a couple weeks, I realized that the problem was caused by both AVG and Avast! Anti-Virus (not running at the same time... come on.) I installed Microsoft Security Essentials as my AV solution and the problem went away. I'm not saying it's not Microsoft's fault, but there IS an easy solution.

Re:Anti-Virus progs

[postmortem]

If you had enabled everyday full scan, then memory cache would be all used up by antivirus, since it would have opened pretty much every file on your computer. Because of this, Windows would slowly put other running apps in background and give more and more RAM to antivirus.

Re:Anti-Virus progs

[Hierophant7]

That makes sense. It also makes sense that roughly 86% of users would enable a full scan every day.

he means the swap file

[snooo53]

dude, you totally missed the point. Even though he said HDD cache he's clearly talking about the performance hit from swapping.

Main memory user

[tokul]

And main memory user is some xpnet-bot.exe process.

User base increases over time.

[tepples]

If you're seeing an actual slowdown in performance, fine, worry about it.

User base increases over time. Even on an intranet server, your company will probably add users when it grows. As your user base increases, you will see slowdowns. If you can catch slowdowns before they happen, you will be more prepared for the DDOS attack that comes when your site gets mentioned in a Slashdot article or when a bunch of new employees go through orientation.

100% CPU usage is a good thing: it means there's a process that's not IO bound.

Or it could mean that you need to optimize the process that uses the most CPU time so that it becomes I/O bound. All other things equal, once all your processes are network bound, you can serve more users.

Multitasking

[tepples]

When you are playing a game, you are not multi-tasking.

Not always. On a PC, you could be running BitTorrent. On a PC or a phone, you could have voice chat in the background. On a video game console, you could be running players 2 through 4 on the same machine as player 1.

Talking about bloat...

[trboyden]

cfwtracker.exe is a 100 KB application that uses 102.4 MB of memory. Someone's program that uses over 100 times its program size in memory consumption probably shouldn't be blasting anyone about memory bloat.

Currently my Windows 7 system with 4 GB of memory only consumes 1.75 GB of memory with 6 different applications open and lots of multi-tasking going on.

Perhaps most of the people on XPnet are running netbooks with 2 GB of RAM? Then perhaps the guy could have a point. However, most Windows 7 ready computers are spec'd with 4 GB of memory, so clearly Microsoft envisions most people will have at least that much memory in their Windows 7 system and this 100% use of memory rant is just hype and damn lies.

XP too.

[Lonewolf666]

Windows 7 seems to place a lot of data in swap, even when gobs of RAM are available. With a large enough page file and Superfetch+Prefetching enabled, physical RAM that's actually in use (not cache!) is usually between 50% and 60% full.

Disable the page file, and the RAM that's in use skyrockets to 70%+ with the same processes running.

I suspect the same on Windows XP, albeit based on different observations.

When I run enough applications to fill most (but not all) of my 2GB RAM on XP, switching between these applications tends to trigger several seconds of disk activity before the newly selected application is in the foreground and fully responsible. I've noticed this when running Day Of Defeat: Source and EVE Online in parallel. Total memory use according to the Windows Task Manager: over 1 GB, but well below the installed amount of 2GB.

    From a well designed memory management and caching system, I'd expect in this situation that the previous background application still sits in memory (one way or the other) and is retrieved quickly without resorting to the hard disk. But obviously that does not happen. Possible conclusions:
1) Task Manager is lying and not reporting some of the actual memory use. So XP really needs to swap.
2) Memory management is poorly designed and swaps needlessly.

Re:XP too.

[bemymonkey]

Windows 7 does this as well. With the page file activated, switching between open programs can take a while, and in these cases there's always a flurry of disk activity.

I'd be interested in a way to turn this off.

non-story clickwhores

[Gothmolly]

Can we raise the intellectual bar somewhat, and not even post these articles? At a quick glance, its a non-issue, AND/OR the authors of the piece ADMIT that they don't really know how to interpret the data.

Turning off the second SODIMM on battery

[tepples]

Using more RAM doesn't use more energy. Either your RAM is powered on, or it's not.

So I have two SODIMMs in a laptop. When I unplug it from the power supply, I want it to free enough caches on the first SODIMM to move all my processes from the second RAM stick, and then switch off the second SODIMM.

Re:Turning off the second SODIMM on battery

[Mad Merlin]

Using more RAM doesn't use more energy. Either your RAM is powered on, or it's not.

So I have two SODIMMs in a laptop. When I unplug it from the power supply, I want it to free enough caches on the first SODIMM to move all my processes from the second RAM stick, and then switch off the second SODIMM.

And I want a pony.

Sadly, neither of these desires make any sense. The extra work your hard drive needs to do with half the RAM will vastly outstrip the tiny power savings you'll gain by using one less SODIMM.

Re:Turning off the second SODIMM on battery

[tepples]

hard drive

My Asus laptop has an SSD instead. Besides, even in a machine with a hard drive, it will still spin up every time a program fsync()s a file after having updated it.

Much Ado About Nothing

[detachable_halo]

My 2 cents: My Windows 7 PC is using the same amount of memory as it did when it was an XP PC, running the same application load. For actual numbers, that was about 45% of 3 GB usable RAM. (out of 4 installed. I just upped it to 8 GB last night because that had been my original design for the PC, and I had gotten my hands on a 64-bit Win7 installation disc; repeat, this upgrade was NOT due to performance issues.)

It's disturbing to me that this company is raising such a noise about this, when they clearly state "We don't know why we're getting these numbers." Well, since we have no idea what's doing it, let's look at a common factor:
>>Users who want to compare their computers to the current WCPI numbers can do so by registering with XPnet and then installing the DMS Clarity Tracker Agent from Devil Mountain's site.

I say we blame it on the DMS Clarity Tracker Agent; when it runs on Windows 7, it causes high memory usage. Let the fear-mongering begin... Who's with me?

Everything but a REGISTER hit is a cache miss.

[Anonymous Coward]

It's pretty bold of you to be telling others that they "obviously don't understand memory access design" when it's clearly you who is clueless.

Your beloved "L1 check" can actually be the second cache miss. The first level of caching in a CPU is its registers. If that cache hit fails, then it resorts to checking the L1 cache, then the L2 cache, and so on.

Re:Everything but a REGISTER hit is a cache miss.

[Anonymous Coward]

...I'm not going to claim to be well versed in modern assembly, but the idea that registers are being used for caching sounds thoroughly wrong to me.

Flawed logic

[damianesteves]

This makes no sense at all: "For the OS to be pushing the hardware limits this quickly is amazing. Windows 7 is not the lean, mean version of Vista that you may think it is." OS Pushing the hardware limits quickly? Really? That's just banter to start some drama. Last I checked hardware gets better with time and OS minimum requirements stay the same. So logically, any OS will push hardware limits the most at the beginning of its life and then as hardware improves, become less taxing.

Available memory != Free memory

[TheLink]

Yeah. I don't have low mem problems with Windows 7. There's stuff I don't like about Windows 7 but "memory hog" is not on the list.

For work I'm using Windows 7 64 bit on a 4GB notebook PC with tons of windows open e.g. a few Explorer windows open, a few Excel "windows"[1], a few Word windows, one Visio doc, Notepad++, Google Chrome, Firefox, putty, Outlook (a resource hog), Communicator, MSN Messenger windows, a Virtual Box Linux vm machine, Microsoft Security Essentials (it's my work PC so it's supposed to have AV) and it typically says 1700 to 2000MB _available_ (depending on how many firefox tabs, how many word docs and virtual machines etc). But overall no mem problem.

And guess which is using the most RAM? Not Virtual Box, not Word, outlook or Excel. It's Firefox with a 173MB working set and 142MB Private Working Set!

Yes it only has 500MB free memory, but so what? The O/S says there's 1700MB available. And so far I haven't had much slowdowns due to low memory issues.

To me the relevant metric for "low on memory" is "Pages Output/sec" (go launch perfmon.msc and add that counter). If that's a constant zero when you or the O/S switches from app to app, window to window, it means it's not swapping out. If it's not swapping out and not getting "out of memory" messages, it's not low in RAM no matter what some random "expert" thinks. And it's zero for me.

The equivalent in Linux for that is the swap "so" column when you run vmstat 1 (or vmstat 2). Same thing there - stuck at zero = not swapping.

I don't think my usage can be considered "light", as it is, what are those users running that's using up so much memory? Symantec or McAfee antivirus? ;).

FWIW, my laptop is not running any of the "OEM crapware" - I did a clean install of Windows 7 months ago when I got the laptop.

If that "expert CTO" can't even give an example of one memory hogging program (or show where Windows 7 itself is using so much memory that it's a problem), then it's likely he's full of crap.

Lastly, it's true my taskbar looks messy with two rows of task buttons, but I don't see the advantage of closing and reopening documents or programs if I'm not running out of RAM yet. I close them if I really do not need them (e.g. the document is out of date and not used for comparison). Otherwise it's much faster to just click a button to show the desired doc, rather than have to reopen it again from scratch (uses less battery power too - except in the case of MS Word which seems to use CPU even when "idle" - haven't figured that one out yet).

[1] By default Excel actually just has one window which changes to display the relevant document depending on which Excel taskbar button you click, whereas Word actually has separate windows for each doc.

Re:Available memory != Free memory

[Anonymous Coward]

If you are running the sidebar you may like this:
http://www.addgadget.com/all_cpu_meter/
It gives you RAM and CPU stats (by core). A good gadget for seeing what the CPU is up to is Top Process:
http://benhollis.net/software/sidebargadgets/

Re:Available memory != Free memory

[TheLink]

If you are running the sidebar you may like to look at this:

http://seclists.org/bugtraq/2007/Sep/134

See the discussion and also the pdf http://www.portcullis-security.com/uplds/Next_Generation_malware.pdf

I'm sticking to perfmon.msc, task manager, resource manager and Process Explorer, depending on the circumstances.

Re:Available memory != Free memory

[jo_ham]

I have no idea about Word itself, but it's not limited to windows - the Mac version of Word also uses noticeable cpu cycles when it is completely idle (ie, far more than any other idle app).

My only thought is some sort of virtual hamster wheel.

Re:Available memory != Free memory

[Anonymous Coward]

Well, that was interesting reading. I just took the time to confirm (by reviewing the css, html, and js for each gadget) that those two make no outside calls for data. There is a boiler plate "check for update" function in both but it is passive, no download option, just the js to render a browser link to the home page. Just for kicks I'll try editing out the update function anyway.

Thanks for confirming my suspicions about gadgets that connect to the net... as tempting as the cartoon and news feeds are I can get that stuff through FF but I do like the stats for my PC available at a glance... and if checking on my RAM and CPU puts me at risk I think I might already be boned anyway because these gadgets are using taskman and resmon as input.

Also thanks for getting me set off on this track. I'm gonna dive into the API and see what kind of goodies I can make for myself.

Re:Available memory != Free memory

[fropenn]

You can open multiple windows in Excel by holding down CTRL and opening Excel from the start menu. This has been really helpful when using Excel on my 2-monitor setup.

Re:Available memory != Free memory

[fractoid]

What I don't get is that they're crapping on about "Windows 7" using a lot of memory but they make no mention at any point of what applications are running. They're not talking about Windows 7 machines sitting there at the login prompt or on a default installation with no apps running, are they?

Re:Available memory != Free memory

[fractoid]

Well, MFC uses 'onidle' handlers which are called continuously while there are no messages in queue. It usually uses these to update UI elements etc.

Re:Available memory != Free memory

[Anonymous Coward]

Yeah that's why I said by default. Holding down CTRL doesn't seem necessary for Excel 2007, just launch another instance from the start menu.

It's still annoying though - can't copy and paste formulas easily, and would be more convenient if you could hold down CTRL and double click the excel file you want to open in a separate process.

4GB of RAM Quad Core Running Win7

[Vamman]

So this article bugs me. The memory usage on my Win7 dev box has been very stable. Win7 uses about 50% of my ram all of the time. This hardware is not exactly new. It was bleeding edge when I had Windows Vista and I admit the memory requirements was a bit higher. Given I've been running Windows 64bit versions since I've started using Quads so I can not report on the 32bit experiences. Time for some upgrades Mr. Chief?

Windows has a bigger swapPINESS

[tepples]

If the RAM is needed for another purpose the blocks can be freed by simply changing a flag and mapping them as quickly as if they had been totally free. This does not slow things down. After a while most of the blocks you need will be in RAM when you need them. Thus caching makes the system faster, not slower.

The above describes Linux. I assume Windows works similarly.

As I understand other comments to this article, Windows runs at the equivalent of 100% swappiness, making the system far more likely to evict a process from RAM than to evict things from disk cache. Writes are slower than reads, especially on a RAID or on a laptop with a low-cost SSD. Or to express it more graphically:

8==D : Swappiness on Linux properly tuned for slow writes
8=========D : Swappiness on a Windows box

There is a new button in "task manager"

[mr_da3m0n]

Since windows vista, you can hit a button in task manager that basically translates to "resource monitor" on my localized system. The "Memory" tab accounts for superfetch and shows a colored bar for memory used by the cache, but that memory is considered "available".

If you'd believe the "physical memory" metrics you'd think out of the 4gb of memory my windows 7 workstation has, only 18 megabytes are available. But that wouldn't be accurate. Right now I seem to actually be using a little over 60% of my memory (a guesstimate based on eyeballing the graph). And I have netbeans loaded with a glassfish instance.

The hard faults column is hardly populated too, which seems to counter the implied claim in the article that superfetch is causing applications to hit the swap more frequently thus reducing performance.

This has been stated many, many times in the comments but just to reiterate with a tangible example: Windows 7 has a better superfetch than Vista. It will cache *anything* you might want loaded in memory at some point based on your usage. For instance if you play World of Warcraft a lot, you'll notice the data files are already in ram when you boot your workstation, ready to be accessed. But again, those files are still marked as "available" so if you suddenly need that memory it will be immediately freed.

So either Barth knows something I don't, or he's an idiot with no understanding of the Windows 7 memory model. Both are equally possible at this point since the article is being mind-numbingly vague.

On a similar note:

[Hurricane78]

I just looked what eats up memory here on my Linux machine.
And the one Java process I have running, eats away 1 GB of VIRT (146 MB RES). Which is a bit mutch on a 2GB machine...
The only thing that comes close, is another virtual machine: Windows XP in Virtualbox. (Can’t run them both together.)
Then Firefox and Thunderbird each take 749/777 MB VIRT (206/134 MB).
After that Amarok (440/38), X (344/205), etc.

And you know what I think?
What the hell do those apps need all that RAM for???
That can’t all be pixel and audio data, can it? And it needs a LOT of parse trees and text to fill up even 100 MB! (Think of a 10MB XML file, to get the picture.)

So, does anyone know a way to find out, what those apps actually use their memory for? Just lots of 64 bit Bool (98.4% waste) values? ;))

Re:On a similar note:

[MikeBabcock]

Your virtual size is not indicative of the size of the program in memory at all.

The resident size is fairly valid, and shows the size of the program currently in memory, but can be both code and data size.

A lot of that is also shared (note the shared field) as its the memory used by glibc or other libraries.

For example, launching four copies of the Java VM doesn't use 4x more RAM. Compile a quick program that just does a sleep for an hour and run a thousand of them. Check the resident size on any one, then add them together, then check your free memory. They won't add up, because a lot of that size is shared space.

Chipset limits on RAM size

[tepples]

now ~10 years later people have squeezed 512 megabytes per stick.

For one thing, a lot of the really old chipsets just don't recognize the larger RAM sticks. I have a PC that came with a 128 MB PC133 stick and one empty slot; I later added a 256 MB (the largest compatible size) for 384 MB, which let me upgrade from Windows 2000 to Windows XP. For another, drivers for 32-bit desktop Windows aren't compatible with the Physical Address Extension needed to access more than 3 GB of RAM, so chipsets for 32-bit CPUs are often limited to 1 GB per slot.

Re:Chipset limits on RAM size

[petermgreen]

Another issue is a lot of slightly older (early C2D generation) chipsets only support a 32 bit address space even though they support both a pair of 4GB modules and 64-bit operating systems.

Comment removed

[account_deleted]

Comment removed based on user account deletion

It's not just memory it's CPU

[Anonymous Coward]

This "System Idle Process" keeps taking up 98% of my CPU.

When is Microsoft going to do something about this pig of a program?

Barth says that Win7 is a resource hog!

[Prod_Deity]

DYAAAH! I heard that! RIP Les Lye

How are they maxed out?

[furby076]

Other then including XP computers in the test pool (which is just plain dumb), how are they getting these numbers? When I bought my computer (two years ago, during the vista days) it came with 8 gigs of ram. Even then it was typical for me to see 4 gigs of ram at best buy, dell, etc. 4 gigs of ram is more then enough to handle vista/win7 smoothly for most users. Most users don't use graphics rendering tools, high end motion graphics, etc. Gaming doesn't even come close to taxing computers

So now I am sitting on a dual core, 8 gigs of ram, and probably average the 15-20% usage range even when playing games like Warhammer Online with the highest sound/graphics settings, while running Ventrilo, Curse, firefox, WoW Mouse software, Logitech G15 keyboard software (with scripting), and other standard stuff (e.g. antivirus, firewall, brother printer/scanner software, and other tools I can't think of right now)...oh and dual monitors

I would say 99% of all software out there does not fully utilize what an "average" computer can dish out - the software is just not that intensive. It reminds me when the SNES came out years ago and they mentioned how much storage capacity the game cartridges had- and what they could handle...then there was an article stating they use less then 10% of the game cartridge and system capacity for the games...so basically the system was 10times more powerful then needed. THere is very little reason for people to buy quad core - because for the most part people can't utilize it to the max even if they are heavy gamers. BTW that tidbit of info I got from 1) a computer sales person who told me not to waste my money on quad core, and instead get dual core when buying a computer from her and 2) my friend who has been in IT for about 25 years dealing with this type of hardware decisions.

More info in TFA

[Anonymous Coward]

The article indicates that they ARE seeing a measurable drop in box performance, and a spike in I/O:

"Other data that Devil Mountain collates as part of a new metric dubbed "Windows Composite Performance Index" (WCPI) quantifies peak processor workload and I/O performance. Both of those measurements are also higher for Windows 7 systems than for XP machines. While 85% of the former are running at peak I/O loads, only 36% of the latter do; the numbers for CPU workload are closer, as 44% of Windows 7 computers are running a computational backlog that delays processing tasks, compared to 36% of the XP systems.

"This is alarming," Barth said of Windows 7 machines' resource consumption. "For the OS to be pushing the hardware limits this quickly is amazing. Windows 7 is not the lean, mean version of Vista that you may think it is.""

Re:More info in TFA

[AVee]

Well, they are seeing a spike in I/O, but they didn't actually measure performance nor did they take into account how people are using their system. So effectively they might just be comparing grandma playing solitaire on her XP machine to a gamer playing some demanding game on his Windows 7 box and claim that Windows 7 needs more memory and I/O. They didn't normalize the results in any way, but you can be fairly certain that the first users upgrading to Windows 7 are the gamers and other more demanding users. Also, Windows 7 machines are more likely to run other more recent software, so perhaps it's just that Office 2007 and the Office 2010 beta are a bigger resource hogs compared to Office XP en Office 2003.

It's all just statistics, but without really knowing what is being measured and blindly assuming that correlation means cause. But hey, we got a percentage so it must be true...

a folder called con

[Anonymous Coward]

I thought Windows just does a land grab of memory on boot. Which is why the VMware hypervisor can share pages between virtual machines, especially empty ones. Beware the balloon driver!!

Seriously though, people here still use Windows and still care what the RAM utilisation is? Colour me boggled.

So How Much RAM for Windows 7?

[JumperCable]

So how much RAM are people suggesting for the average home & basic business users for the 32 bit edition & the 64 bit edition?

Re:So How Much RAM for Windows 7?

[Anonymous Coward]

32 = 2
64 = 4

Re:So How Much RAM for Windows 7?

[snikulin]

32 = 3
64 = 8

Re:So How Much RAM for Windows 7?

[kcbnac]

New system? I recommend 64-bit, as it is where everything is going. Office 2010 will be 64-bit native, might as well keep up with the times. No cost difference, and lets the OS use more for whatever. 32-bit only makes sense if the system won't support 64-bit, or has a stupidly low memory limit - non-upgradable memory, for example.

32-bit = 4GB
(You're building a new machine, 2GB sticks are a decent density. Windows will use leftover stuff for cache, and RAM is cheap enough)

64-bit = 4GB
(You're building a 64-bit machine, 4GB is a good basis - and if it isn't enough later, you can double that)

RAM is the fastest way to make a PC run smoother (along with optimizing the config, nuking useless apps, etc). Might as well toss a nice 4GB at it, then it will be fast for the life of the machine. Games *have* in the past brought 2GB systems to their knees (Hellgate: London would eat 2GB for breakfast, and they had a 64-bit client; since then I just say 4GB as a minimum, that way the OS has plenty of RAM to work with outside the game - man did it swap with only 2GB...until they fixed the memory leak, but even then 2GB was pushing it)

The only I learned from this article

[Anonymous Coward]

is that the people at Devil Mountain Software's community-based Exo.performance.network (XPnet), dont really seem to know very much about OSs or how they allocate memory.

how to allocate memory

[Anonymous Coward]

What specifically was inaccurate in the provided metrics. Do you have different numbers, if so please provide citations.

Bullroar

[Scarumanga]

So not true, probably applies to machines with less than 4gb of ram. I have 8gb of ram in my machine, i usual have around ~5gb free at all times. So this is total bullshit. No one should have less than 4gb in a new PC these days anyways. There's a lot of dumb folks out there who only have 1gb of ram and then they complain that their computer is so slow, when 1gb was the ideal amount 6-7 years ago, but this is 2010, ram is cheap, people gotta get with the program.

It's like the matrix...

[flappinbooger]

When Mr. Reagan went back into the matrix, he's sitting there eating the steak with the agent, and he says something like "I know the steak isn't real, but I don't care"

With windows 7, the perceived performance boost over vista may not be "real" but .... I dont care! (And nor does anyone else, except maybe those people whose laptop batteries bite the dust once they put on 7)

Using what you paid for

[ramzafl]

Windows 7 actually makes use of the RAM you paid for, NEWS AT 11!

No kidding.

[endus]

Installed x64 on my 4gb machine and the performance was just ridiculously bad. I am a photographer and do a lot of image editing...couldn't even keep Cap One, Photoshop and iTunes open at the same time...especially if 7 was trying to thumbnail images in a folder (thumbnailing is broken and ridiculously resource intensive despite Microsoft's claims that its more robust in this OS). Noticed that it was swapping to the disk like crazy and ordered another 4gb. Definitely much better now, though I think 12gb would not be totally out of line for x64 with heavy applications. I haven't even TRIED to edit HD video yet...that is not a prospect I am looking forward to.

I pity people running the x86 version of the OS that are maxing out at 4gb. Definitely buy 64 bit (even though its more of a beta than a real OS) if you do anything memory intensive.

The one good thing about all this is that HOPEFULLY...FINALLY...maybe this will push Microsoft to push 64 bit more. They need to abandon 32 bit and force application writers and hardware manufacturers to start making 64 bit native applications. Working in a medical environment BELIEVE me I understand the need for backwards compatability, but the fact is that the resources are just not being put in to 64 bit to make it a really viable platform and even moderate power users are going to start bumping up against the 4gb limit.

Yea, I read that thing saying that the 4gb limit is a product-based limit rather than a technical limit but either way...it appears that x64 is where MS is choosing to support > 4gb so lets get serious about it.

Sorry, only works under linux

[SmallFurryCreature]

On Linux if you know you got enough ram and can life with the risk that if a program goes memory wild you are going to be in trouble, then you can turn swapping off. The problem on my own linux machine is that torrenting and large file serving confuses the machine, it tries to cache these file reads but this is rarely needed. So if it starts to swap, it is swapping stuff it will never need again.

It is fine that it caches files in memory, since this is fast, but caching stuff on the HD that it just read from that same HD and will never need again... that is pointless.

In theory, windows should be able to do the same, but its page file has become so much a part of the system that it seems to need it, even if it has tons of free memory. I had a 32bit game on a 12gig machine and it STILL insisted on a page file WHY?

So, turn of swap on linux if you got enough memory (and make sure you do because opera/firefox with plenty of tabs open suck up memory like you wouldn't believe) but on windows, you can't.

Re:Sorry, only works under linux

[Unequivocal]

I think I've had a few cases with a xen'ed instance of Linux on a shared host where running out of ram caused the OS to simply cease to function. I don't know what the settings are that cause this but since the host doesn't provide for a swap space we have to keep everything within our allotment of RAM or poof. It's a good measure of our code though - whether we are efficient with ram usage, though painful at times.

Not sure if there would be an equivalent on a home machine for Linux. My experience with Windows (xp and earlier) is that if you run without a disk swap, windows will usually pop up nice little messages saying "out of ram" -- but I think I recall some times on older versions (NT) where we'd get blue screens if we were running intense processes like active SQL servers without disk swap.

History repeats itself once again

[God_TM]

Isn't this technology ('supercache' or whatever you wish to call it) carried over from Vista?

I thought it was big news back when Vista first came out...

http://www.codinghorror.com/blog/2006/09/why-does-vista-use-all-my-memory.html
http://forums.overclockers.com.au/showthread.php?t=640296

Re:History repeats itself once again

[God_TM]

My bad... it's called SuperFetch, and here's what MicroSoft has to say about it:
http://blogs.technet.com/askperf/archive/2007/03/29/windows-vista-superfetch-readyboost.aspx
and here is tomshardware's analysis of it:
http://www.tomshardware.com/reviews/windows-vista-superfetch-and-readyboostanalyzed,1532.html

GrandParent is +1 informative. So is parent

[kirillian]

It's nice to see a couple people who paid attention in either their college virtual memory course or got some real life experience. Memory usage is, quite often, a useless statistic without other qualifying statistics to help interpret it. The first thing that I do when checking out a computer with problems is add page faults to the visible columns in the task manager. I don't think the article demonstrates that the TSA either has the knowledge or use of the proper statistics for making the conclusions they made. While possible that they may not have released all the data they have or something similar, it definitely sounds like an instance of someone spouting off their mouth before acquiring the data they need to draw reasonable conclusions and/or incompetence at some level. The fact that their data could not determine whether the memory usage was due to OS or application usage is also another sign that they may be jumping to conclusions.

You can't look at the page fault counter

[rabtech]

In Windows, when a program loads, all executable code is immediately mapped into the process address space, but may not actually be loaded into memory yet - those mapped pages have their "swapped to disk" flag set. They just happen to live in the original DLL instead of the page file... just because a page lives on disk doesn't necessarily mean it lives in the page file. Prefetch keeps stats about what pages from what DLLs are actually used most of the time and it pre-loads them into RAM, but the rest of the pages will stay on disk until you need them. All of this stuff shows up as page faults because technically they are, but it doesn't mean you are swapping to disk.

Windows will also not overwrite free'd pages immediately; if they are not allocated and get touched, it just flips the bit back to put the page in use. Windows 7 also greatly increased the caching mechanism, which aims to put 90% of all free RAM to use as cache, the majority of that is read cache so it can be thrown away at a moment's notice if an application needs it. Again, all this activity makes the page fault count tick upward.

Bottom line (as others have pointed out): This is a total non-story.

Is anyone surprized?

[Anonymous Coward]

Q1a: What is the Microsoft recommended RAM for Windows XP?
A1a: 64MB, 128 recommended.
Q1b: What is the average RAM installed in the test pool?
A1b: (a guess) between 768MB and 1GB.
Analysis: People using 6-8 times the amount of recommended (up to 16x the minimum) RAM ony experience saturation of RAM about 45% of the time when tested.

Q2a: What is the Microsoft recommended RAM for Win7?
A2a: 1GB (32bit) 2GB(64bit) minimum. (recommended not specified), additional 1GB for XP mode.
Q2b: What is the average RAM installed in the test pool?
A2b: (a guess) Less than 4 GB certainly, probably close to if not under 2GB.
Analysis: People using close to and up to double the MINIMUM RAM size experience saturation about 87% of the time.

Microsoft's requirements don;t take into account additional application requirements, running services, user settings, and more, however, it;s pretty clear that if you have 8 times the recommended memory, you;ll hit the cap less often than someone with double it, especially when the new minimum is 8 times the previous recommendation!

If you're not using at least 4GB in Win7, plus readyboost, you probably don't have enough RAM unless you're a Netbook class user load (web/email only, little or no multitasking, no graphics, small screen resolution). Any heavy users of Win 7 should be seriously considdering 64bit home premium, which supports 16GB. Pro supports 192.

Isn't it a good thing that it uses 100% of RAM?

[Blakey Rat]

I don't see how this is a bad thing... using 100% of RAM doesn't necessarily mean that the system is relying on the disk more than if, say, 50% of RAM was consumed... if the thing you need is on the disk, you gotta hit the disk, right? However, if RAM is 100% full at all times, there's a much greater chance that the thing you need is already in RAM, making it quicker to retrieve and therefore making your computer faster.

I think the thing people miss on this debate is that RAM can be emptied *instantly*. It's not like you have to spend 4 seconds "draining off" the mis-used RAM before you can load in the thing you need, right? So using 100% of RAM can't possibly make your computer slower, it can only make it faster.

I'd actually be more alarmed if your RAM *wasn't* 100% full at all times. I'd be asking my OS, "hey, why aren't you caching more?"

Re:Isn't it a good thing that it uses 100% of RAM?

[Cassini2]

Using 100% of your RAM can easily make your computer slower. Firstly, if a page is "dirty", it needs to be saved to disk before it can be used for another activity. This causes a write to disk before a read, automatically making the read slower. To combat this, sometimes the data is automatically written to disk when the computer is idle, thus creating a pool of non-dirty pages. Under Windows XP, the pool of non-dirty pages includes data from previous disk reads (the read disk cache), and pages of program memory (virtual memory). The problem is "which page do you overwrite next?" If a program is about to read from a portion of its program memory, or is about to read data sitting in the disk cache, you don't want to delete that page. It is just that the operating system doesn't know which piece of data the program will need next.

The result is that having empty RAM is almost always beneficial. Blindly swapping program memory to disk is not a good idea. It never really was. RAM is much less expensive than it used to be. For many problems, current computers have vast amounts of RAM for the algorithms and software they implement.

On today's computers, a memory leak can effectively stop/bring down the computer. The real function of the page file is to soften this problem. Instead of dying relatively quickly, today's computers will slow to a crawl, and then die. The page file causes this.

If you have good, well behaved software, and 4 GB of RAM, you will likely experience better system performance under Windows XP by turning the page file off.

Re:Isn't it a good thing that it uses 100% of RAM?

[Blakey Rat]

Ok that's retarded.

Using 100% of your RAM can easily make your computer slower. Firstly, if a page is "dirty", it needs to be saved to disk before it can be used for another activity. This causes a write to disk before a read, automatically making the read slower.

While that's true, it doesn't matter because those pages would be in-use no matter how much RAM you have.

The "free" RAM (that is, RAM not being used for applications, that is, RAM pages that won't ever be marked dirty) is all caching-- disk and DLL caching. It'll never get marked dirty, and therefore never need saving to disk before being reused.

The problem is "which page do you overwrite next?" If a program is about to read from a portion of its program memory, or is about to read data sitting in the disk cache, you don't want to delete that page. It is just that the operating system doesn't know which piece of data the program will need next.

That's... somewhat true.

First of all, while the OS doesn't know what piece of data the program will need next, it certainly can (and does) make a guess based on what the program has requested in the past. Most programs allocate most of their memory when first starting up, and the OS is going to be able to predict these bits almost 100% of the time. (Unless it's the first run of that program. But it's not magic.)

Secondly, even if the guesses are bad, even if only one guess in a thousand is correct you're still better-off. Tossing away RAM from a bad guess takes zero time, for all practical purposes. In reality it takes time, but it takes less time than the disk takes to load up the data to fill that page, so in practice it takes no time.

Therefore, the penalty for a wrong guess is zero. The reward for a correct guess is significantly higher than zero. It's a win-win situation.

The result is that having empty RAM is almost always beneficial.

Wrong, for the reasons listed above.

Blindly swapping program memory to disk is not a good idea. It never really was.

True, but Windows doesn't do that and that's not what we're talking about. That would never happen unless RAM was 100% full of application data. Even if that was the situation, it's hardly "blind" swapping, as the same heuristics that allow Windows to predict what an application will load next can help it predict what an application will never touch again.

But to repeat: this isn't the situation we're talking about. We're talking about a computer filling unused RAM with disk and DLL cache. You're talking about a computer whose RAM is 100% full of application memory. Apples to oranges.

On today's computers, a memory leak can effectively stop/bring down the computer. The real function of the page file is to soften this problem. Instead of dying relatively quickly, today's computers will slow to a crawl, and then die. The page file causes this.

Ok... and? What's your point? Are you saying that if we didn't cache to RAM this wouldn't happen?

If you have good, well behaved software, and 4 GB of RAM, you will likely experience better system performance under Windows XP by turning the page file off.

Unless that software wants to memory-map a large file, in which case your performance will go straight to hell. Also, in practice, there's a ton of stuff Windows loads initially that can be (more or less) swapped-out immediately and stay there.

For example, the login screen is likely to only be displayed once, ever. And when it's needed again, there's plenty of time to load it-- why leave it in RAM? (And Windows is smart enough not to.)

The only time I'd consider turning VM off entirely is if you're running MacOS 8. Which had worse than useless virtual memory support. You're not, you're using an OS made in the 21st century, turn on the fucking swap file.

That's supposed to happen.

[Anonymous Coward]

Isn't RAM there to be used?
Doesn't Seven cache commonly used stuff in RAM?
Can't this simply be a case of people using their computers to full potential?
Memory efficiency and light footprints are good and all, but, I don't know about you guys, I buy RAM (and CPU/GPU power and disk storage, etc) with the intent of using it. If memory isn't being used, it's being wasted. It's one thing to use up memory when the computer is idle (even then, caching commonly accessed stuff in RAM is an efficient use of otherwise unused RAM, it's a feature, and it's supposed to do that), ot's another matter entirely when your memory is maxing out during actual use.

I never quite got where this memtality that computer resources are sacred and shouldn't be used up came from, not in this day and age where hardware is dirt cheap anyway. I get it, in the old days where resources were both scarce and expensive (but even then, it existed to be used).

Well, you've got to keep them upgrading hardware..

[Anonymous Coward]

Not that this upsets me: use Linux and it is nice to be able to get a machine with 4 gigs of memory for $400. I just think it is pathetic that an operating system uses over 512 megs of memory. Just think about all the things you could do with 512 megs of memory 10 years ago. I wonder how much of that bloat is DRM?

Tuning the Windows Kernel...

[TemporalBeing]

Even in Windows 95 there were registry settings to change the behavior of whether the kernel would use all the physical memory before swapping pages to disk. It actually did quite well to improve performance unless you used a dial-up modem with Outlook.

That said, Windows has historically always kept physical RAM usage low and paging to disk high. However, if I recall correctly they may have changed that for Windows 7 - using more of the physical RAM before paging to disk. That is probably what is being observed in the reports, so it's likely a very unfair comparison of apples and oranges.

You can turn the pagefile off, but know your apps

[Joe U]

The system/OS wil be fine without swapping. It's your applications that may suffer.

It depends on what you're doing, if you're reading email all day and have 4GB of RAM, you can turn your pagefile off with no ill effects. My 2GB netbook runs without a pagefile and has a 128MB ramdisk as well. It's optimized for reading email and slashdot and never writing to the (amazingly slow writing) SSD unless absolutely necessary. If I decide to install a word processor, I might be in a little trouble.

Windows XP was released in 2001

[Anonymous Coward]

Yes, the 9 year old version of Windows uses less RAM than the latest, shiniest version of Windows! Since we're comparing apples to oranges, why don't we take it one step further and complain about how XP uses more RAM than Win95 or NT (which great ran on 8MB RAM).

Srsly, get over it.

WTF

[pclminion]

If every single last byte of RAM isn't filled with something, then RAM is being wasted. You don't waste RAM by using it. You waste RAM by not using it. This is like buying a bigger desk so you can put more stuff on it, then claiming that by covering your desk in stuff you are "wasting" the desk. What the fuck? If you have 2 gigs of RAM and you only want 1 gig to be in use, the solution is to remove that extra gig of RAM. It sure won't be "wasted" sitting on the shelf, now, will it?

(If the system is swapping madly, then there is a problem. It means that less important data is being held in RAM in preference to more important data. But that's a separate issue. The mere fact that 100% of physical RAM is in use, is a GOOD THING.)

Re:WTF

[Anonymous Coward]

Just like RAM, covering your bigger desk means you have to move your clutter in case you want to put something on it that's not there already.

Seriously, what's wrong with loading on demand, rather than Windows trying to guess (incorrectly in many cases) what you might need? Despite what people think, Windows is not psychic.

Re:WTF

[pclminion]

Seriously, what's wrong with loading on demand, rather than Windows trying to guess (incorrectly in many cases) what you might need? Despite what people think, Windows is not psychic.

If you load some random set of pages into memory, then there is a non-zero probability that one of those pages will subsequently be used. If you load NO pages into memory, there is ZERO probability of having a useful page in memory. Obviously, if RAM is already full, you should not be speculatively loading pages. But if free pages exist, and the current IO pressure is low, it causes no harm to pre-load pages, especially if you are good at guessing what will be needed. It's IO pressure, not memory pressure, that matters, because it takes essentially zero time to kick a page out of RAM. On the other hand, queueing up a bunch of read operations that might be unnecessary will make it harder to quickly load a truly useful page when needed.

From the article, it looks like somebody measured how much physical RAM was in use, saw that pretty much all of it was, and thought there was something wrong with that. That's dumb. Was the system swapping?

Re:WTF

[Anonymous Coward]

Is there any way to disable this behavior in any OS, period? I know you can get rid of Superfetch in Windows (which, by the way, loads programs FASTER with it disabled and makes my system MUCH more responsive -- and I have oodles of RAM), but is there a way to completely turn this behavior off? I'd love to see if it actually makes things even faster, completely contrary to popular belief just like Superfetch.

F'd up

[Anonymous Coward]

You know you really F'ed up when you got 90% of Slashdot standing up for Microsoft side of the story...

Umm, sure.

[jonadab]

As best I can tell, at least 95% of "Windows 7" PCs actually have Windows XP installed (typically through the OEM downgrade option and/or volume licensing, although some users just get out their old XP disks and install that way).

Are there computers out there that actually *run* Windows 7, on a day-to-day basis? I haven't seen them.

You're Crazy!

[czehp]

According to my memory monitor, I'm only consuming 15% of the RAM I have available and I'm running Windows 7! Of course it may also be because I have 12GB of RAM...

Windows still playing catch-up ...

[lwriemen]

... to OS/2. I remember when the Windows FUDsters used to post comparisons of memory usage all over the OS/2 newsgroups, because OS/2 was loading DLLs into RAM and using all the memory. I guess what comes around goes around. :-D

Not my results

[gravis777]

On neither my laptop, which has two gig of ram, or my desktop, which has 6 gig of ram, have I ran out of RAM with Windows 7. In fact, my experience is that Windows 7 uses LESS ram than Vista did (don't have actual numbers to give, sorry). That being said, I do USE the RAM on both - I do HD video editing on the desktop, and I multitask the crap out of my computers. Its not unusual for me to be rendering video, have uTorrent running, and acting as a media server, while I am Photoshopping stuff (on top of the RAM, thank goodness for multi-core processors and the ability to change thread priority in my rendering software). Truthfully, no one I know who is running Windows 7 has yet to hit a place where they "ran out of ram" - even those with only a gig (then again, those people tend to run instant messanger and a webbrowser most of the time, not exactly memory intensive applications). In my experience, most people this day and age have at least a basic understanding of numbers such as mHz and Ram, and know what they are going to be using a computer for, and can generally explain their needs to the computer salesmen. Even your most basic system on the market nowdays has enough processor power / memory to run Windows 7, a web browser, maybe an e-mail program, an instant messanger program or two, and possibly a word processor, if needed (unless you are at a computer store that is trying to unload old stock on you.). If you are running out of memory, its because you either running an older computer and installed a new OS on it (in which case, you probably need more of an upgrade than just a new OS), you did not properly convey your computing needs to your computer sales person (when people suddenly decide they are going to start running Photoshop and do video editing and playing games on their base systems), or you constantly work on large rendering projects (Photoshop, Premiere, Maya, etc) where no amount of ram is ever going to be enough for you.

The principle is sound.

[argent]

Not even Microsoft gets things wrong 100% of the time.

I am not sure that the actual algorithm and heuristics that Vista/Seven use is ideal... there are reports of cached applications causing page-outs of running programs working set... but the principle of using "free" memory as a cache is completely sound.

Windows uses virtual memory no matter what

[sonciwind]

I've found Windows uses virtual memory no matter how much memory you have available. That's why I max out the memory on my notebook and turn off virtual memory. But Windows has continuously become so much more inneficient that it really seems intentional rather than just lazy which I originally thought was the reason.

windows 7 and memory...

[chentiangemalc]

I really don't believe this is the normal trend for Winodws 7. I've been using Windows 7 on four machines - 2 running x64 and 2 running x86 (2 laptops, 2 workstations) since beta in January and haven't seen memory usage problems. I've also tested numerous other hardware models, many on or below the recommended spec - i.e. systems with 512 MB ram, and also machines with 1 GB ram, In the performance tab of task manager in Windows 7 you will see large amount of ram used for cache. Windows 7 is very responsive, and fast...basicially totally awesome, I couldn't stand having to go back to XP. Possibly there are some applications people running that are memory hogs, but it's not Windows 7.

Check your own software?

[Nuvious]

I have had 3 different machines all running Windows 7. One is a netbook with 3g's of ram, another is a ZOTAC Ion system with 2G's of ram and finally an HP Tablet with 2G's of ram. I have NEVER seen 90% RAM usage and I'm constantly running Eclipse, several browsers, audio streamers, and other applications. Maybe XPNet should be checking their code for problems with their software and not throwing the blame at Windows 7.

Don't forget crapware

[Ryzzen]

Nearly every computer you buy in store (or online even) comes preloaded with so much junk, it wouldn't surprise me if your typical computer boots with 75% of actual memory used up. Just walk into a Best Buy and try using one of their display models. Your average Joe Schmoe doesn't know that all those little icons in the system tray shouldn't be there, much less how to disable them. This isn't Win7's fault, it's the OEMs'. Agreed, this is a very lousy report with way too many variables.

It is the choice of terms, seriously

[Ilgaz]

On OS X, user launches Activity Monitor, sees "Active" "Inactive" "Used" and "Free". All marked with easy to understand colours like red/green/yellow.

Windows user (including me, when I first saw Vista) sees something horrifying like "8MB" of Free RAM. If he/she doesn't know the very complex inner workings and strategy of OS, he will sure say "OMG damn memory wasting OS".

If a person monitoring his OS one way or another since 1980s get tricked by that "Free RAM" indicator, you can imagine the rest. It is the choice of words/layout.

Also that real stupid, lame and archaic "page file reserve" system. MS had a really good explanation to debunk thousands of old pages claiming user will have better performance if he sets a dedicated page file (usually gigantic) and we were all convinced. What happened and why do they actually do it in core OS level now? Fragmentation of a file which already carries random data at random pointers is an issue now?

Windows Ready-Boost flash cache feature

[billstewart]

Windows Vista or 7 can use a variety of kinds of flash memory as disk cache. It's the kind of thing you might as well do - fast flash is pretty cheap, and cheap flash is really really cheap, and while the transfer rate isn't as fast as rotating disks, avoiding rotational latency is a huge win these days, plus (unless you're running a your systems with lots of disk drives) you get to use it in parallel with your disk drives, so your applications and data don't have to compete with each other for disk rotations or bandwidth.

Windoze is shit?

[dogzdik]

Well yeah - I could have told you that. The worlds never ending consumer driven beta test for products that are not fit for market....

Comment removed

[account_deleted]

Comment removed based on user account deletion

Windows 7 _IS_ mean and lean.

[Sunnz]

What? Windows 7 _IS_ indeed the mean and lean version of Vista. It is is just still not as fast and feature rich as other modern operating systems, you can cut the fat of a pile of crap all you want, but in the end crap are still crap.

Really??

[swilkers808]

I run Windows 7 Ultimate on my Lenovo netbook and only use 35% of my ram (3 gigs). Maybe its time for 86% of the people to upgrade hardware.

Calling BS

[dark_requiem]

I've been using 7 since the beta, and have seen absolutely no evidence to support this. I have 8GB RAM, and swap disabled. Running the OS, three VMs (1GB assigned to each), a couple firefox sessions with 20-40 tabs each, a game of Civ4, and an HD movie, I won't come close to maxing out my memory. Divide the available memory in half to 4GB, and assume that most people aren't running VMs with 3+GB memory usage, and the picture painted in TFA appears highly unlikely. Unless they're gauging systems running Win7 with 2GB memory, I call BS. And if they are gauging systems running Win7 with 2GB, then who the hell cares? You can't be surprised when you're running an OS on a machine that barely meets minimum specs and you start to max out the hardware.

Yet Another Rebuttal from xpnet.com

[Anonymous Coward]

Apparently, they've posted another rebuttal:

http://exo-blog.blogspot.com/2010/02/editorial-what-took-you-so-long.html

AC