Slashdot Mirror
Ask Slashdot: Are Post-Install Windows Slowdowns Inevitable?
blackest_k writes: I recently reinstalled Windows 7 Home on a laptop. A factory restore (minus the shovelware), all the Windows updates, and it was reasonably snappy. Four weeks later it's running like a slug, and now 34 more updates to install. The system is clear of malware (there are very few additional programs other than chrome browser). It appears that Windows slows down Windows! Has anyone benchmarked Windows 7 as installed and then again as updated? Even better has anybody identified any Windows update that put the slug into sluggish?
Related: an anonymous reader asks: Our organization's PCs are growing ever slower, with direct hard-drive encryption in place, and with anti-malware scans running ever more frequently. The security team says that SSDs are the only solution, but the org won't approve SSD purchases. It seems most disk scanning could take place after hours and/or under a lower CPU priority, but the security team doesn't care about optimization, summarily blaming sluggishness on lack of SSDs. Are they blowing smoke?
The security team runs the scans during the daytime because that's when everybody's laptop is powered on and connected to the network. Too many people shut off their machines at night, or carry their laptops home, so the scans won't reliably run if they do them then.
There is probably some kind of creative, adaptive scheduling solution that could fix this, but their management software might not have that kind of support.
The problem is when you start installing other programs. In your post you mentioned installing anti-malware. Every time you open a file it takes 10s to 100s of milliseconds to scan it. The problem is exacerbated if you have an antivirus program, then every file gets scanned twice. If you want preformance then turn off both and do scans when the user isn't working on their machine.
Ask Slashdot: Are Post-Install Windows Slowdowns Inevitable?
When patches and updates together end up being larger than the original [OEM] install, you can see why the slowdown is inevitable.
Sounds easy to see why. No?
I can't help it, and maybe it's my imagination and perception bias, but to me it seems to be that as soon as a new version of Windows is approaching or even out the door, the old version starts to slow down considerably. And like clockwork you can rely on MS themselves and various testers claiming (of course with good benchmark proof) that the new Windows is so much faster than the old one was.
The rational person in me would say that after a bazillion patches and service packs, the stitched together hodgepodge is of course crawling along because there's a lot of dead weight being lugged around and worked around.
The tinfoil hat enthusiast in me on the other hand claims that it's deliberate to make the new Windows look better despite being essentially the same.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
1. Your operating system is very old at nearly 7 years. Time flies bye and I laugh at the companies who are angry at the prospect of starting a WIndows 10 migration acting somehow that 7 just came out last year and is all so new etc. The point is you will have 200 updates and the .net framework will need to re-compile to your cpu dependent architecture each time an update hits for better performance. Have fun with that one.
2,
Windows ROT is soo last decade with WindowsXP.
It is caused by poorly written programs that run as admin and write to the registry each time they run. So you run the app 200 days a year and it creates 200 forks of the registry that need to launch in parallel at startup :-)
With UAC WIndows 7 doesn't have this problem.
3. Do you own an Samsung EVO SSD?
If so they will slow to a crawl very rapidly without a patch. They will hang after a few months of heavy use for several seconds before a file even transfers. I only buy the pro drives. Go google this up as their is an engineering flaw which impacts the read due to the way the cells are manufactured.?
http://saveie6.com/
I disagree.
Windows 7 is still very snappy on my AMD phenom II. XP had WIndows ROT problems with poorly written apps and even updates which forked the registry many times which impaled the startup process. I have not seen a slowdown at all. I do admit I upgraded to an i7 and now have 8.1 on it but I occasionally use the other system.
I think he has .net framework recompiles going which happen after these updates are one of those defective evo 840 drives which will halt after a few months without a patch to fix the charge leakage bug.
http://saveie6.com/
I had this problem all the time with XP but I have not noticed it since I installed Windows 7 over 5 years ago on my current PC. The only problem I have now is that the WinSxS folder is gigantic, likely due to all the Windows Update patches over time. My poor SSD. Windows 10 claims to use "3GB" as a minimal requirement, we'll see how that holds up (I expect not well at all) but Windows 8/8.1 supposedly cut down on used disk space by the OS a bit so I'm optimistic for now.
Define "worse." As someone who dual-boots Windows 7 Home and Linux Mint 17.1 Rebecca, I find Mint works much "better" in terms of responsiveness. It boots faster, shuts down faster, opens programs faster, runs quieter, etc. Sure, it requires more expertise, has less proprietary software options, and obtains the performance improvement, at least in part, by sacrificing certain bells and whistles, but I've gone from a 90-10 Windows-Linux time split to 10-90 as I've grown more comfortable with the latter.
In my experience, Linux desktop response suffers way more heavily under high disk load then Windows desktop response. Something with the way Gnome and KDE are prioritized in the kernel loop I would expect. Run something in the background that is chewing up the disk and expect windows to draw very slowly.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Windows machines in recent years have become extremely bottlenecked by drive performance, especially in the case of laptops which are so popular in companies. Laptop hard drives are slow, capable of only about 80 IOPS which is about the same speed they were 10 years ago, whereas mainstream SSDs by comparison, can typically deliver 80,000 IOPS. Since once you get Windows loaded up with all it's random messy software it's disk access ends up being tons of tiny reads, IOPS is a much more important number than transfer rate, and SSDs are literally 1000x faster. It can mean the difference between a 20 minute operation and one that takes a few seconds.
If you are in any way in control over your corporate purchases, never *ever* buy another laptop without a SSD. It's false efficiency, wasting very expensive time to save a relatively cheap expense. 256GB SSDs are under $100 and will handle most corporate work just fine. Up to 1TB, the expense is almost negligible and it will pay for itself almost immediately. Your IT department will be happier, your workers will be happier, your machines will be more secure because scanning them is a lot less intrusive and can happen more often. Your IT department should have a pile of SSDs ready to be deployed into any machine that needs to be re-imaged or where the user needs the speed. Not doing so is wasting money.
> I recently reinstalled Windows 7 Home on a laptop. A factory restore (minus the shovelware), all the Windows updates
No you didn't. You *thought* you installed all the updates because Windows lied to you and said you had. Windows Update has a horrible habit of checking to see what updates are available **for the state of your machine right now** and then telling you that it's done installing updates when those are installed, when in truth there are pending updates that required previous updates to be installed before they could subsequently be installed that Windows Update won't tell you about until you re-discover what updates are available. After an install, force re-scan after every reboot to see what new updates are now available and when you reboot and re-scan and it says you are done, you are actually done.
set softtabstop=4 shiftwidth=4 expandtab nocp worlddomination
Not my experience. It used to be the case in Windows 98 but I haven't found my systems to be slowing ever since I bought an SSD. SSDs solved all of my problems and they're rediculously affordable.
I also have a cluster of windows machines performing raytracing and other extremely performance driven tasks--I can't tell the age of an install based on performance.
This is all just superstition at this point without numbers. Yes if you install a third party anti-virus solution and you have a bunch of auto-installers running in the background your computer will run "Slower" than it did without anything running in the background but that's not Windows' fault and that's true of every operating system regardless if it's *nix or Win*.
Actually having experience. Oh and the fact that I just set up a new windows 7 VM and from the fresh install on the DVD and how it ran, compared to after applying all updates it lost all of it's speed.
Nothing installed but windows updates. on the exact same hardware. Absolute solid proof to me.
Do not look at laser with remaining good eye.
things like this have been said about windows for decades. It's never been true. I know because I've had operational business machines at each version of windows running for over ten years each.
These types of problems happen *and are henced resolved with brand new way-more-powerful hardware) when multiple components aren't spec'd together.
Any given component has many bins. You can get any cpu at six levels of l3 cache, for example. Drives can be 5'400, 7'200, etc.
The trick is not to get the most possible performance (which is akin to buying a new machine a few years later). The trick is to match the performances across the various components, so a single component doesn't become the bottleneck.
Especially because some components, when acting as the bottleneck, can create serious slow-downs. Often actually making something else SLOWER will make the over-all machine much faster.
An over-simplified example is that a slow hard drive can create disk-thrashing scenarios -- one of the worst slow-downs common across the board. But a slow cpu will remain slow and steady, and never wind up thrashing the disk.
Learn to balance the vital components of a system, and it'll stay consistent for a decade.
(this was written on my 8-year-old vista machine, still working, still business, still gaming, still full-speed)
One way that windows 7 (in particular) slows down, comes from the use of the winSXS folder.
Basically, because the windows software ecosystem is so... Plagued.. with legacy software that expect older versions of system libraries, Microsoft invented a solution to detect those dependencies and satisfy them with those older libaries in a sandbox-- the WinSXS folder.
As time passes, and updates happen, system libraries get updated-- instead of being replaced, they get moved to the winsxs folder and archived. This is so when your bitchy internal-only legacy application that is oh-so-mission-critical that it simply cant be rewritten for a modern OS gets run, it can continue to run.
The downside is that as this treasure trove of old libraries grows, the penalty of the checking routine becomes more and more apparent. (also, it consumes more and more disk space.)
Other forms of slowdown are not specific to windows 7 and newer however.
The registry is a binary file that must be parsed to find entries inside it, and it too can become fragmented. As changes are CONSTANTLY happening to the registry, the (actual) structure of the registry can become more and more byzantine. Since such changes are completely unavoidable with daily use, the slow degradation of this system is also unavoidable unless you boot from a golden image each and every time. This has been a problem since at least the 9x days. Back then, you could automate registry defragmentation with a bootup script because of the complete lack of filesystem security on FAT-- (Tell regedit to dump the registry in its totality into an exported text file, then tell it to rebuild the registry from scratch using that text file dump, then cleanup the temporary files afterwards.) You cant do that with modern flavors of windows because 1) you cant invoke scripts that easily on bootup anymore 2) the registry files are protected with NTFS security descriptors, 3) the OS locks the registry basically as soon as NTLDR finishes, so you cant replace the registry files while live.
There are of course, the other causes of slowdown that come from cumulative misconfigurations that happen from automated updates, but meh.
There's no way around the sludge that is the Windows registry, or f.e. Windows' tendency to regularly enter a heart-stopping "drive frenzy" for no apparent good reason, but it doesn't all have to end up with the familiar ugly crawl we all know friends' and family's Windows boxes come to just months after freshly installed. Next to me is a Win 7/64 machine whose current installation is close to 5 years old, and has seen thorough use (as with all my Windows machines during the 15 years I've used it on personal level) but it's still quite snappy and acceptably fluent even if slightly more sluggish than when the installation was new. The only difference is that I take care of my personal computing, and avoid the pitfalls that "computer illiterates" so often fall for.
I have 5 machines running Windows 7 and this is not a problem, but then again I am not using encryption, just standard antivirus software. The ones with SSD boot drives are faster, but none of them has issues with running slow. In fact, they are quite responsive. The oldest one has had Windows 7 running on it for over 5 years without slowdowns or problems and the control panel says that it has 163 programs installed.
A brain is a terrible thing to waste... Mind? That's debatable.
Well, it could be the scheduler you're using. If I'm not mistaken, the linux kernel by default uses CFQ. I use BFQ and from the benchmarks it pretty much beats CFQ hands down. I also use LXDE, which is more lightweight than KDE. To be honest I don't notice that much of a difference between Windows and Linux, but that may be because I use a SSD.
Wish I had mod points for that post. Run Process Explorer and TcpView to see what is going on.
Dear Slashdot, I am a self proclaimed computer expert. Windows seems slow to me. Give me reasons to install Linux even though I can barely operate Windows.
Only the State obtains its revenue by coercion. - Murray Rothbard
The security team is right. SSDs are the single biggest performance improvement you can add to a computer (even an old computer). If your company is upgrading computers after they get 5-7 years old, but refusing to buy SSDs, they're wasting money. In particular, if they're upgrading management's high-end machines while the low-end machines are still being used by the rank and file, they're doing it completely backwards.
The problem is most people focus on the high-end numbers. How many GHz does the CPU run at? How many MHz does the DDR3 memory run at? Improving the high end doesn't help as much to improve productivity. It's already fast, meaning you're waiting a very small time for it to finish. Making it twice as fast just means the very small wait period shrank a tiny amount and is now twice as small.
If you're serious about improving performance, you get the biggest return by upgrading the slowest components. The slowest part of a modern PC is the HDD. When reading small files (not sequential reads, which really come into play only when copying large media files from one drive to another), they max out at about 1 MB/s. In contrast, the next slowest component - system RAM - is currently on the order of 10 GB/s. In other words, in terms of wait times a 1% improvement in HDD speed will have the same impact as a 100x increase in RAM speed. Now, consider than a SSD will get you at least a 30x improvement in read speeds for small files (about 30 MB/s seems to be average) and there is absolutely nothing you can do with the RAM or CPU which comes anywhere close to the amount of time you'll save by replacing the HDD with a SSD.
If you've got old computers, you should be upgrading them with a SSD instead of replacing them with new computers (with a HDD). Continue to use the old computers + SSD for a few more years, then upgrade them and transfer the SSDs to the new computers. The only exception is if the computer is so old you can't install enough RAM to run modern applications. (Another rare exception would be Northwood and Prescott-era P4 CPUs, which burn so much electricity you'll actually make back the cost of upgrading them via lower electricity bills in a couple years.)
On top of that, SSDs can actually look up small files faster than the computer can request them. So if you've got a virus scan running on a SSD, you can continue using the computer like normal with almost no impact on performance. In fact I usually run my weekly virus and two malware deep scans simultaneously on my SSD laptop, and I can still use it for web browsing or office tasks. When a virus scan runs on a HDD, the HDD has to spend all its time reading files the scan is requesting. As a result anything you try to do with the computer which requests data off the HDD will bog down.
One might be puzzled by my response, but I say no because technically anything can be fixed, the only question is how.
I've been struggling with this issue lately myself as my own laptop (which is not underpowered by any means) has been experiencing incredibly slow login times for the Windows 7 install I have on my HDD. I also have an install on my SDD, but aside from bootup the performance difference is negligible for me(I also use it a lot less so it doesn't have all my software installed). The hard drive in this case is a 2TB Samsung Spinpoint M9T at 5400 RPM. Slower RPM, but it's a super dense 2.5" laptop drive.
I've made some progress in speeding it up, especially the login time which was atrocious... Removed an update that caused some Windows crap to be re-verified or something all the time, removed several things from startup and switched non-essential services to automatic. Eventually I did get the logon process to not be too bad and Windows would become responsive after maybe 40 seconds instead of 5-10 minutes. It's still not as fast as I'd like, but it's much improved.
But the problem with this is that I'm shooting in the dark and have to rely on trying pretty much every suggestion on the web there is. And here is the difference between my Windows installs and my Linux installs. GNU/Linux is open source, virtually everything you use in it is. The system is also designed to be tinkered with and the bootup processes are all opened up for any level of configuration that you desire. You can screw with your init system, the kernel itself, your bootloader, anything... So with the sources to all these pieces, I think figuring out what's wrong is relatively easy.
Come Windows, everything is closed source. The problem can be fixed, but you're stuck with decompiling and trying to debug perhaps even the kernel itself if you want to solve any problems. How are you going to profile bootup or login times? Can you easily find a sink for disk or CPU usage in certain functions in the Windows source code? Probably not. It's really challenging to figure out what's going wrong in this case. The best I can hope for is to look to people who have gotten a lucky guess or someone who is so absolutely hardcore that they've debugged a closed source operating system.
Just my 2 cents.
I almost can't believe that people are still asking questions about this, but I suppose I'll have to have to let this one go off easy.
Windows is not responsible for doing this to itself. It is your anti-virus software that is doing this + assuming you aren't one of those people that downloads and installs every program trial and freeware from the net.
Try it. Uninstall your anti-virus software completely.
Back in the Windows XP days when Vista was released I figured out what caused all the performance issues and it was the anti-virus software. I had reinstalled Windows XP many times so I was quite familiar with its snappy and responsive performance on a new install vs one year later. After uninstalling the anti-virus software, everything was precisely as snappy and responsive as it was on Day #1. It still holds true today. The effect is lessened significantly if your system runs on SSD (anti-virus know nothing on Day #1 of their install, but over time they learn the system and gradually bring the system to its knees).
I can translate from KMH to feet per minute. I know that seconds and even jiffy exist, but what is this 'snappy' and 'sluggish' you talk about.
Most likely you are now used to the speed and now just want it to go even faster.
First: do a complete new install and see if it is still 'snappy'. If it isn't, then it is your perception of the speed. If it is, start adding things as you lost likely did and see when it starts to happen.
I must say, my PCs are just as fast as they were when I got them out of the box, or at least almost. Yes, I have added software that will slow things a little bit down, but only if I measure it, not when I actually use it. At least, I am unable to notice the difference.
So please come back with information like: when I reinstalled Windows 7, the time to load a 17MB image into GIMP was x time. After 4 weeks, the time it takes is X+Y. I have only done upgrades and between upgrades A7 and A8 I noticed a time increse in the loading if the same image from 0 to +Y.
The update did change FileA.exe and FileB.exe. What I further did was ...
And that is how you do a technical posting. Not "I think it might be, like, you know, sluggish like a slug, not, I don't know,. like snappy, like a snapper."
Don't fight for your country, if your country does not fight for you.
What slows Windows down are not the updates. You can have a Windows server running for years, installing updates and never slowing down.
You user profile is what is slowing your Windows down (the content of c:\users\%USERNAME%), and NO you can't just delete it... try to login with a new user and you'll see. It has been like that since Win95 and Microsoft has never fixed this.
From many years of working with Windows PC's there's one thing I know for sure and that's that one of the major reasons for Windows to slow down over time is the default setting of the virtual memory paging file which is "Automatically manage paging file size". As the page file expands and contracts on this setting the file gets ever more fragmented and access to it gets slower. When I first setup a new computer (with Windows pre-installed) one of the first things I do is change that setting from automatic to a custom size and make the initial and maximum size the same so hopefully it's allocated all in one piece and as close to the beginning of the disk as possible where access is fastest. If a computer has been running for years on "Automatically manage" it's page file many be in thousands of pieces and that could possibly slow the computer significantly when the page file is used. There was a utility called PageDefrag for Windows XP that allowed you to defragment your pagefile but the author Mark Russinovich never updated it to work with newer versions of Windows so there is no easy way to defragment a pagefile on Windows Vista and up but one method I've used with success is to use a partition manager to reduce the size of the boot partition (pushing it farther along the drive) and create a small block of space (perhaps 40 to 60gb) in between the system reserved partition and the boot / Windows partition; after that format it and give it a dive letter like X: and then put the page file there. When you do that it's as close to the beginning of the drive as possible and at a static size Windows never has to work to expand or shrink it and it never gets fragmented.
One other thing is that the author mentions Windows 7, at the end of 2014 over about a three month period I built eight new computers for people who wanted quality hardware (all eight were identical in motherboard, CPU, RAM and hard drive) and seven of them I installed with Windows 8.1 and one the person requested Windows 7; I noticed during installation and in general using the computer with Windows 7 that it was noticeably slower than the computers with Windows 8.1 so Windows 8 appears to be faster than Windows 7 on the same hardware, at least that's my observation. (and that's Windows 7 x64 versus Windows 8.1 x64)
Another thing that slows computers down is the accumulation of temporary files, there's a tool someone recommended to me called TFC (temp file cleaner), you can find it here http://www.geekstogo.com/forum... and it really does a phenomenal job; many computers that I've used it on show marked speed improvement after running it.
Running Windows 8 on non-SSDs, I just found performance went up incredibly when I turned off the superfetch service. There's some sort of bug where it gets to 100% disk usage after a while if you're not restarting every day or two. (Sleep isn't enough). Slows the whole damn system down and task manager and resource monitor just show that you're using the pagefile, making it tricky to track down.
It might not be a problem with SSDs, which have very different read characteristics.
I can't help it, and maybe it's my imagination and perception bias, but to me it seems to be that as soon as a new version of Windows is approaching or even out the door, the old version starts to slow down considerably.
Correlation is not causation. The guys at MS are professional engineers--they may have different philosophies or coding styles or project priorities than you do, but they're not slowing things down in order to make you buy the next product. You're much more likely to run into that with a local guy or a disreputable company. And you might not like MS, but they haven't been a disreputable company for decades. Even if they had an inclination to be (and they don't), they're too big in the business-to-business space to risk their reputation.
What happens is your systems get slower as they get older, other systems get faster, you install more stuff, your drives fragment a bit, you add extra hardware, maybe you get malware you don't know about, etc...
We have a similar scan problem, but our co's policy is to not shut down PC's at night so that they can get Windows updates. But the scanning still happens during the day even if one leaves it on.
Couldn't a scan rule be put in place that only scans during the day IF the night scan didn't complete? Anybody know of a tool like that for McAfee? Does McAfee have a scripting language or scheduling rule engine? Or, a 3rd party add-on?
That way ONLY those who turn it off at night get "punished" by sluggishness. (Or if a Windows update interrupts an anti-virus scan, which may happen from time to time, but that's better than always day-scanning.)
McAfee could make a nice profit even by selling such a rule tool. It's like being paid to create a problem and being paid again to solve it: Kinda like Congress :-)
Table-ized A.I.