Slashdot Mirror
Bulky System Requirements for Windows Vista
unsurreal writes ""A Tech Strategist within Microsoft, Nigel Page, has gone on record to discuss the hardware requirements for Windows Vista, due out next Christmas." The next year is going to be an interesting one as hardware vendors smile towards the shocking new recommended hardware needed for the next generation Windows operating system." From the article: "Graphics: Vista has changed from using the CPU to display bitmaps on the screen to using the GPU to render vectors. This means the entire display model in Vista has changed. To render the screen in the GPU requires an awful lot of memory to do optimally - 256MB is a happy medium, but you'll actually see benefit from more. Microsoft believes that you're going to see the amount of video memory being shipped on cards hurtle up when Vista ships." Coverage available at Tom's Hardware as well, with a semi-transcript at Tech Ed.
Windows XP Professional: 128 megabytes of RAM or higher recommended
Windows Vista: 2 Gigabytes of RAM recommended
WTF??
I missed that all-important adjective "video". Never mind.
"I'd rather be a lightning rod than a seismometer." -Ken Kesey
Except that they are removing the GDI functionality. All GDI calls will just act as special wrapers back to the vector display calls. MS is not making this OS to be a simple upgrade from XP. They started from scratch and they are compartmentalizing or outright removing a lot of legacy stuff (which is good, it leads to better design overall). The GDI is one such module that has been removed.
Space for rent, inquire within
While the PowerPC chips in the Xbox 360 may be similar in instruction set to the G5, the chips are VERY different. It uses only in-order instruction execution, and not out-of-order, which has been standard(for powerful CPU's) since at least the Pentium Pro(?) era, excepting the Itanium, which has the compiler do the OOO scheduling.
The XBox 360 has 3 very small and rather simple PowerPC cores, and the Cell uses 1 such core, and the 7(?) SPU's along with it.
Does anybody have comparisons for OS X machines? They render much of their OS in the GPU, right?
A lot of the lower end machines still ship with 32 Megs on the card, and run fine (provided you've got a decent amount of system memory). Obviously that's too low for serious gaming, but the OS has no troubles with that amount of memory on the GPU.
Having 256 on the GPU would be on the extreme high-end (only the highest end powermac ships by default with a card that big), not "a happy medium" for OS X.
I'm typing this from my Vista beta install on a 3-year old Dell Dimension 4400, P4 1.7GHz, 512MB RAM and a Matrox P750 VGA card. Hardly a high-end PC these days. Even this first beta, it's been running well so far, does a lot better on suspend/resume than XP did for me and doesn't seem sluggish. Sure you'll be able to get more bells and whistles up and running on faster hardware, but I have no complaints this far..
Before you flame me for being a MS zealot, the Vista machine is next to my Slackware 10.1 box and my really old Pentium 166 that is installing SCO OpenServer 5.0.2 as I type this. Computers are fun, regardless of the OS they run..
To Terminate, or not to Terminate, that's the question - SCSIROB
For a rate of their beefing up...
Windows XP: 128MB RAM, 300MHz
Windows 2000: 64MB RAM, 133MHz
Windows 98: 16MB, 66 MHz
Windows 95: 4MB RAM, 386 or higher
I looked for some older requirements, but it's a good start, and shows approximately the equivalent of solid state advances etc. Yes, they beef it up, but fairly on par with new tech.
Microsoft Sucks, F/OSS Rocks. I get mod points now right?
Well, in comparison, MacOS X 10.4 (which most definitely DOES use the GPU for a lot of the graphics work) required 256MB of system RAM - and a massive 16MB of graphics RAM!
h tml
http://www.apple.com/macosx/upgrade/requirements.
Which does raise the question as to what the hell Microsoft are doing that means they require the same amount of graphics RAM as MacOS X needs for the system!
People should not be afraid of their governments - Governments should be afraid of their people.
Windows 95(mostly) and NT were written from scratch. To a point so was 2000 (based off of NT). But XP was not written from scratch, it was an upgrade from 2000. Vista is not based on XP, it was a total rewrite. Why else would it be taking over 4 years of development work? If it was just a new UI and a few changes under the hood, it should have been out years ago. The only reason it could be taking so long is either a) they are idiots and can't program or b) they did a drastic rewrite of the whole OS from ground up.
Space for rent, inquire within
SATA NCQ does *NOT* give SCSI performance.
This is not to say it's not a hell of a lot more useful than not being able to do disconnected writes at all, but pre-insertion of write barriers instead of post insertion via scheduling is really a poor-man's version of I/O concurrency.
Unless you go out of your way to do a FUA (Force Unit Access), on SATA, there is no guarantee that write data has been committed to stable storage, rather than just cache.
In SCSI tagged command queueing, you can be guaranteed that the write has been committed to stable storage before the write is acknowledges as completed (yes, it's optional to turn this off in mode page 2, but only idiots do it).
The upshot of this is that the OS must issue FUA on writes and stall the pipeline for other writes that don't require a commitment to stable storage (e.g. FUA for metadata and journalling, no-FUA for other data).
This is (effectively) the difference between DOW (Delayed Ordered Writes) and SU (Soft Updates), which is what makes SU so much more effective than DOW.
Further, it means that the OS can't use the acknowledgement to schedule future operations on the disk, without knowing ahead of time the FUA is necessary for a given write.
The issue here is that if I'm, for example, updating the contents in a single directory entry block on disk in two different processes, instead of deciding to delay the second update until I know the first one has completed (via the acknowledgement), I must issue the first one as an FUA command, and then the second one as an FUA command, which adds latency to my pipeline.
"Mr. SATA, I've worked with Mr. SCSI, and you're no Mr. SCSI".
-- Terry
A 32-bit integer takes up the same amount of memory in a 64-bit system as in a 32-bit system. Just because your processor width has doubled does not mean you've automatically upsized all your variables. ASCII characters do not suddenly require 16-bits each, nor has Unicode ballooned to 32-bits/character. In short, the data in the database continues to occupy the same amount of bits as before -- not double.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
a little from column a a little from column b
with apologizies to Matt Groening
Why do people keep talking about Vista requiring a minimum of 512MB of RAM and a video card with 256MB of video memeory? I've seen beta 1 running on a laptop with an intel video card (16MB) and 256MB of main ram! While it didn't run optimally, it did run okay.
This is like the craziest urban myth ever! Perpetuated by geeks the world over. Geeez! Unless of course Beta 2 is a major departure from Beta 1. Get your hands on Beta 1, if you care, and have a look everyone.
What if they release Vista, and nobody bought?
As if people have a choice. If you go to a computer store in 2007, every computer will have Vista preinstalled. (Except the Macs.)
What is the compelling reason to upgrade to Vista?
It doesn't matter, since most Windows sales come from new machines.
This is an inaccurate post in certain ways I am afraid.
You cannot meaningfully talk about a "median" hardware platform over a 6-10 year time scale. That's at least 2 upgrade cycles in duration.
When I ran XP the first time I checked to see how much RAM it was using and it was over 64MB without anything else loaded up.
That's wasn't criminal on its release date and it doesn't tally with (assuming the hyperbole of reports is accurate) Vista requiring high-end hardware on release.
I am also baffled as to how Vista's alleged requirement for powerful hardware is at all 'foresight'. Of course it'll run better on faster hardware, everything does.
Here's a prediction - Vista will not require the massive resources people are fearing to run that well. At least you'll be able to buy a reasonably-priced PC that can cope fine.
Anything else would be commercially inept.
I performed a test installation of the Vista Beta 1 (build 5112) on a Dell C640 Latitude laptop, which is equipped with a modest Mobility Radeon 7500C and 16MB graphics memory, and 256MB system RAM. I didn't do benchmark tests, but I can say that although the installation took almost FOREVER (seriously, I drove home, went to lunch, came back and it was still nowhere near complete) and the installation media was HUGE, the resulting ghost image itself was only 1.1GB compared to a base XP ghost image of half that size which I don't think was too terrible in the disk space department. The OS itself ran only a little slower than XP SP2 does under those hardware limitations. There were noticeable lags, but it functioned as well as I would expect anything Microsoft related to function on limited specs. I personally think the new interfaces are cute, but doesn't hold a candle to aqua or enlightenment, etc. I work for a corporation with a little under 30,000 users and the word from the boss is that we are not going to go to a Windows Vista image (which means, unless they get screwed into having to).
The latest iBook has a Radeon 9550 with only 32 meg of RAM. However, it can use all of the Core Image functionality.
I agree with you...it's a shocking comparison. OSX ran on ibooks which had ATI Radeon cards with 16 meg of memory, if I recall correctly. OSX looks far better than any Longhorn/Vista screenshot I have seen. What the hell are Microsoft doing that requires 16x the video memory and looks worse?
Nice, clean ANALOG RGB signals MUST be presented to the CRT cathodes before the tube can present an image. And there are beautiful horizontal and vertical sync signals available at the deflection yoke.
And if you break the airtight seal, the monitor won't do an HDCP handshake anymore.
The AC gets *most* of the issue.
...and they all pretty much cost the same as the GX1's that we bought to upgrade people to back at Y2K.
Most large companies are on 3-year PC refresh cycles already anyway. 3 years is a reasonable time to depreciate them off their books, and it's exactly how long Dell's extended warranty lasts on them.
When you want "a new PC" for an employee at any sufficiently large company, you place a request for the standard model. That standard model is "whatever Dell is selling this month, bundled with a license for whatever OS Microsoft is selling this year."
In our case, it just moved from the GX280 to the GX620. Next year, it'll be a machine that runs Vista just fine, and it'll cost almost exactly the same as the 280's from last quarter and the 620's from this quater.
It does when Microsoft implements the 64bit OS.
64bit _code_ is usually 15% larger than 32bit, and I'd expect the larger address pointers to require comparable increase in the amount of memory for data structures.
- These characters were randomly selected.
The native integer type will stay 32 bit. AMD64 is an LP64 architecture, so longs and pointers will be 64 bits, ints stay 32 bit. And longs getting bigger really doesn't matter, people using longs for anything besides pointer math are being dumb anyways, and should fix their code. Having 64 bit longs just allows people who need 64 bit integers to have them without resorting to the slow "long long".
So really, its just pointers doubling in size that should effect your memory usage. This will not do anything remotely close to doubling the memory usage of an OS. We've had 64 bit architectures and OSs for years, you can look at them to see what kind of memory requirement increase to expect.
Which does raise the question as to what the hell Microsoft are doing that means they require the same amount of graphics RAM as MacOS X needs for the system!
To be fair, Microsoft doesn't require a 256MB video card for Windows Vista. The requirement will probably be similar to Mac OS X: a video card that can display the resolution you want, at the colour depth you want. That's it.
The 256MB figure is for the new eye candy, and not just that, it's for the new eye candy to run at full speed and not start chugging or such. Mac OS X will also perform a lot smoother, the better video hardware you throw at it. Is 256MB excessive? Probably. But not a requirement.
I was just PC-shopping, and saw quite a few $600 PC's. While most under-$1000 PC's use on-board Intel graphics, there's a big 'gotcha' in the $600 range: No AGP or PCI-Express slot!
A lot of people buying in this range could get a nasty surprise if they find they have a need for a better graphics-controller.
but in the Java world (where I work), about half of memory typically contains pointers
Ok, I know you put "typically" in there, but it really depends on what you're doing. If you have multi-megabyte data structures (eg in a large cache to reduce db traffic) then I'd be very surprised if you had that many pointers relative to actual data. Of course, it depends on the structure of your data structures...
it's probably easier for them to prereq 2GB of ram than 1.4GB.
Possibly, but it's also easier and more purchaser-friendly to say 1.5GB rather than 2GB.
Bottom line though is that I simply don't believe the requirements. XP32 runs fine in 128MB of RAM. Sure, you wouldn't want to run Doom 3 on that, but the OS is fine. I refuse to believe that Vista64 is going to require 16x the RAM. If nothing else, the PC market simply isn't ready for home machines that are that close to maxing out their RAM.
(Typical "32bit motherboards" support up to around 3-4GB of RAM; I don't know about ones for 64bit CPUs, but at this end of the market I can't see it being much higher)
It's official. Most of you are morons.
It wouldn't actually double the amount of memory used, but padding for byte alignment can eat up quite a lot. Depends on how your code is structured, a program with lots of small variables would show this effect much more strongly than one that deals with large arrays (or carves off a hunk of memory and manages it manually).
I'm always surprised by the number of coders that will declare variables as short to "save memory". Declaring variables at less than word size is strictly useful for disk/wire compatibility, and then only with the appropriate pack pragmas.
Quartz Extreme doesn't totally shift rendering to the video card, only compositing.
The 3d scene that is an OS X desktop today is comprised of a relatively small number of surfaces, i think like one or two for each window, maybe each doc icon, each menu that pops up, and so forth. The contents of each of those are for the most part still drawn with the CPU into a backing store by the drawing subsystem Quartz 2D (or Quickdraw if your app is rockin' old school).
In old OS X, the "Quartz Compositor" (don't confuse with "Quartz Composer" a new app in 10.4) ran in a CPU process, compositng the backing stores to the screen with transparency effects, transformations, drop shadows, etc. Quartz Extreme is the change that shifts this compositing program onto the GPU instead. Now, for example, transprency and transformations effects that could not be done for OpenGL windows or video (which bypass the backing store system, and so the old compositor), now all works great and with no CPU hit.
Developed for OS X 10.4 is something called "Quartz 2D Extreme" (don't confuse with plain "Quartz Extreme") which pushes more of the rendering work onto the GPU (and probably moving all of the backing stores into VRAM), which probably makes it more equivalent to Longhorn new model. I believe even in 10.4.2, it still cannot by switched on (i wonder why, hmm, perhaps it needs more VRAM than available on today's machines?).
no sig, no plan, no clue