32bit Win7 Vs. Vista Vs. XP 641
An anonymous reader writes "ZDNet's Adrian Kingsley-Hughes tested the latest Win7 build against XP and Vista and came to a surprising conclusion: Win7 performs better than the other 2 OSs in the vast majority of the 23 tasks tested. Even installation. 'Rather than publish a series of benchmark results for the three operating systems (something which Microsoft frowns upon for beta builds, not to mention the fact that the final numbers only really matter for the release candidate and RTM builds), I've decided to put Windows 7, Vista and XP head-to-head in a series of real-world tests...'" This review shows only a 1-2-3 ranking for each test, so there's no sense of the quantitative level of improvement.
I question the results. (Score:4, Informative)
Take results with a grain of salt. He ranks Vista as better than XP on the AMD machine and as nearly equal on the Pentium machine.
Of course, the AMD machine has 4 GB of RAM and the Pentium machine has 1 GB, so that could have something to do with it.
Re:Still making 32 bit? (Score:5, Informative)
I agree. Nobody is selling 32-bit processors anymore.
Linux can handle 32-bit applications on 64-bit OSes. Surely MS can do the same?
Re:Still making 32 bit? (Score:5, Informative)
32bit or 64bit is essentially meaningless...
Unless you have more than 3.5 GB of RAM
Re:I question the results. (Score:4, Informative)
It's possible that the people who compiled the test results rated the OS's from 1 to 3 with 3 being the best
Microsoft has a good version of Vista! (Score:4, Informative)
Their 64 bit version of Vista is actually the best consumer level OS they've done so far. It's the version that should become Windows 7. It's stable, fast (way faster than the 32 bit version on my machine), and its backwards compatible with almost every application that I've tried.
If they made the default install 64 bits, they'd actually be pushing forward an improvement in their consumer OS. As it is, we'll be living with Vista mk. II.
I'll bet the folks who work on the 64 bit version are scratching their heads wondering why they bother!
Re:I question the results. (Score:4, Informative)
i've found out that generally speaking ZDNet articles are total bullshit, with no relevance to the real world.
This article and your example is just one example of that.
Re:Still making 32 bit? (Score:5, Informative)
lol. you've drunk the kool-aid, 32bit or 64bit is essentially meaningless
There is kool-aid, but you need to check you own cup.
If you are referring to the Apple marketing machine, they ya, 32bit and 64bit are not much different, just larger memory addressing. (Of course OS X is still a 32bit OS could be the reason they like to create this mis-perception.)
On a real 64bit OS, there are 64bit registers and tons of other tricks and optimizations that happen, let alone full 64bit drivers that can shove data to devices oh like Video cards much faster.
If you look at Vista x64 it performs 15% faster than Vista x32 if you have 2GB of RAM.
This includes not only the OS's operation, but even 32bit applications running on the OS.
You see when you have a 64bit memory addressing and can optimize for this in the memory manager you no longer have FS and pagefile lookkup tables for extended amounts of RAM.
You also can do like Vista x64 does and shove two 32bit memory writes into on 64bit address space, so when it can, you get double the read/write performance out of the memory chip because you are pulling two 32bit chunks in one read cycle.
And we could go on and on and on...
Understand yet?
Re:win7 performance (Score:4, Informative)
Who likes chairs anyway?
Steve Ballmer, that's who.
Re:How does it "feel"? (Score:4, Informative)
How often do you really have to move 100 MB or 2.5 GB of files around?
A benchmark like this still probably matters though, as if it's fast on moving 100 MB (a size more easily measurable than 10 MB), it's likely faster at 10 MB too. And it's at these ranges it starts creeping into everyday use and the "feel" you're talking of.
Re:I question the results. (Score:5, Informative)
Take results with a grain of salt. He ranks Vista as better than XP on the AMD machine and as nearly equal on the Pentium machine
Sadly, as much as the SlashDot world not like to believe, this is accurate.
If you have 1GB of RAM even on old hardware, Vista is as fast as XP, as the extra RAM offsets the Vista features overhead and Superfetch and other tricks of Vista help make up performance gains.
With 2GB of RAM, Vista will be faster, even if you have a 800mzh PIII and a 1998 ATI video card.
Vista or should we say the NT kernel in Vista is not slow or bloated, it is the extra features that Vista is doing that consumes RAM that offsets its performance gains over XP. (Search Engine, etc.)
The CPU cycles for the Vista features are light, it is all about RAM. Just like with virtually every Windows and known OS update in history, they want more RAM for the features they add.
- Even for Leopard to perform as fast as Tiger you need 1GB of RAM, which is funny considering Apple was making fun of Vista for the exact same reason.
Here is how it works:
512MB RAM - XP > Vista
1GB RAM - XP = Vista
1.5GB+ RAM - Vista > XP
Windows7 so far is showing that even on 512MB is faster than XP in many cases, which is the result of the event based service manager, that unloads processes/services when not needed and saves RAM.
An example on a running test system with 3Ghz P4 and 1GB RAM:
Vista 41% - OS Consumed RAM
Win7 20% - OS Consumed RAM
See how that might help the Vista RAM overhead and put Win7 back in line with XP?
PS And on this test system Vista is faster than XP - even in gaming with a Geforce 5600 video card.
Re:win7 rocks (Score:4, Informative)
So, you're saying that when Win7 nags you and clicking the nag opens up http://www.microsoft.com/windows/antivirus-partners/windows-7.aspx [microsoft.com], they're pointing you to uncertified software? BTW - I just went to his system and did the install again and didn't get any warning about installing uncertified software, so I'm guessing it's signed.
Are you guys actively testing Win7, or just ragging on people that don't report the bestus experience ever?
Re:Still making 32 bit? (Score:3, Informative)
While it may take a particularly clever compiler to use the extra width of the registers when operating on 32bit data, even the most basic compilers will be able to take advantage of the fact that there are twice as many general purpose registers.
Re:Still making 32 bit? (Score:4, Informative)
Unless you have more than 3.5 GB of RAM
Unless you allocate more than 3.5 GB per process.
PAE has gotten around the 4 gig limit a long time ago.
Re:Still making 32 bit? (Score:5, Informative)
Despite supporting PAE, Vista-32 still limits addressable physical memory to 4 GB (Wikipedia [wikipedia.org]). PAE will also run into problems at 64 GB, whereas 64-bit machines shouldn't reach another addressing limitation until they hit 16 EB.
Transitioning to 64-bit is a better solution in the long term.
Re:Still making 32 bit? (Score:3, Informative)
Re:I question the results. (Score:5, Informative)
Unless you need the proprietary ATI or nVidia drivers, one reboot at the end of installation and it's done. And, if you do need to download those drivers, that's only one more reboot. Two at most, and you're done.
Not true, even if you use [gxk]dm, you should be able to "activate" the new driver (after updating xorg.conf) by killing the dm. It'll auto-restart and thus load nvidia.ko.
Of course, God only Smiles on you if you use startx.
Re:I question the results. (Score:5, Informative)
Re:I question the results. (Score:2, Informative)
What exactly discredits my words? Vista got so bad at audio performance that they had to hack around it by slowing down other parts of the system (by adding 'multimedia class scheduler'). It exactly mirrors my observations.
I had to add a workaround to my application because waveOutGetDevCaps function became about 100x slower on Vista than on XP so it became impossible to use it in a single thread with interactive audio.
Also, empirical observation: it became impossible to work with video on Vista in VMWare on my notebook. Even though it worked fine on XP.
As for gaming benchmarks: I don't care about them. There are too many different parameters affecting them.
Re:I question the results. (Score:4, Informative)
Re:I question the results. (Score:3, Informative)
With a fresh partition he meant on a single clean 100% disk coverage partition before installation of the OS. As in not one partition for each OS. I know it looked weird but ..
Re:I question the results. (Score:5, Informative)
That depends on which driver and how you install it in the first place. I use Fedora, so I can only use that for my example. If you use a different distro, YMMV. If you download the nVidia driver from the OEM site and install it, you will have to reinstall it every time you update the kernel, because of the way it works. Or, you can download kmod-nvidia and install that, because that gets updated whenever the kernel does. And, just in case there's a time gap, you can also install akmod-nvidia. That checks on boot to see if you have the latest kmod, and if it's out of date, builds another one on the fly.
So, the answer is, yes, you do have to rebuild/reinstall modules, but the process can, and often is, done either by the distro maintainers, or on the fly without any user intervention.
Re:I question the results. (Score:2, Informative)
Don't be coy. When you say MS had to resort to some sort of hack to make certain interactive processes more reliable, you have a certain connotation. Yet those "hacks" have been going int the Linux kernel for over a decade. Are all the different CPU schedulers and their ways of detecting which processes are interactive a "hack" in the same connotation as you refer to the NT Kernel?
Come on...
And on top of all this, you're complaining about audio performance in a VIRTUAL MACHINE. Surprise: VMWare drivers aren't as fast, as efficient as the real thing. Additionally, to support more features than they should, they often lie about their capabilities and end up overtasked. Many of the earlier virtual NIC drivers for many virtual machines would choke on gigabit traffic. Shocking, I know, that software emulation of a hardware interface can be slower.
Try testing your interactive Windows code... in an actual Windows installation? Virtual machines add a whole layer of obfuscation to your already flimsy anecdote and, frankly, makes you Just Another Troll.
Re:I question the results. (Score:3, Informative)
I quite understand what Vista audio system does (down to the driver level). For one thing, it doesn't use hardware acceleration (!!!) anymore and does everything in software.
"It is by far the most advanced personal computer audio system available on any platform." - is a complete lie.
JACK (http://jackaudio.org/) is probably the best personal high-quality audio system (it has a zero-latency design). It's followed by PulseAudio which is now not quite yet zero-latency but much more efficient.
Adding some more latency into audiobuffers to adjust timing is a fairly trivial task. Also, a good implementation would just turn off this misfeature if the system uses only one sound sink.
Re:The devil is in the details (Score:5, Informative)
How about this.
HP DV9825NR
1.83 GHz T5550 Intel
4GB DDR-800
320GB SATA
512MB GeForce 8600M GS
RealTek HD Audio
I had to hack drivers to get the video card to be seen under XP.
Used for audio production, I made a quick multi-tracked setup using CoolEdit under both Vista and XP, then tested mixdown/encoding from .WAV to MP3.
XP beat Vista - 13 seconds in XP vs 28 seconds in Vista, for the same minute and a half of music.
For gaming, even with my hacked driver to get the video card recognized, playing Fallout 3 in Vista at 1280x720, medium details, gives me an average of 32 FPS. In XP, same detail settings and resolution, I average 40, following the same path, same difficulty. In XP I also lose the stuttering issue in Fallout 3 that Vista users seem to be getting, which seems to be caused by the audio subsystem, as turning audio acceleration to Basic stops about 90% of the crashes, and fixes several noise loop issues.
So, Vista SUCKS. My laptop is dual-booted with it and XP, and I only use the Vista partition for internet stuff, webcam, skype audio chat, etc. Games and any WORK gets done in XP.
I want to try 7 on this laptop.
Re:Still making 32 bit? (Score:1, Informative)
I don't profess to have a knowledge intricate enough to claim SnowLeopard is 100% 64-bit, not to mention that Snow Leopard isn't actually out yet and things may change. What I have learned is best represented by the graphic on this page [appleinsider.com] where there is an end-to-end path of 64-bitness for Snow Leopard that wasn't there for previous iterations of the OS. To me that means a 64-bit OS. Perhaps to you it doesn't.
You claim it doesn't matter for the windows world and then pick out a counterexample for an application written in Carbon. You're either deliberately obfuscating the issue or not understanding what's going on.
1. If you write your application in Cocoa a 64-bit version is a recompile away just as in Windows
2. Apple never sold Adobe anything, the devtools are free
3. What never got moved to 64-bit was the Carbon framework which was more of an interim measure to support software written for the PowerPC days and Adobe, understandably, were reluctant to do a complete re-write (using, I believe, Codewarrior)
Perfectly. I don't disagree at all. Of course a lot of this was necessitated by the transition to OS X and from the PowerPC architecture to x86.
Yes, why do you keep asking me that?
I don't quite get what you mean by 'noble' company. Apple certainly don't need me to defend them and I do remember the ads, though I remember it as the first 64-bit laptop (though that was also false). Yes, they were misleading, but Apple's certainly not alone in that game. They did a similar one for the FASTEST PC on the planet...
I'm not disputing windows was there first*, what I'm saying is that while Microsoft has its feet in both 32 and 64bit OSs Apple is trying to move the entire product line. If Snow Leopard delivers what is being promised Apple will not be able to claim beating Windows to 64-bit. What it can claim is to be the first away from 32-bit*.
Understand? ;-)
* for comparisons between Apple and Microsoft at least
Re:I question the results. (Score:3, Informative)
True. You have to calibrate it first. But once calibrated it delivers to the position you told it your head would be in.
Re:Still making 32 bit? (Score:3, Informative)
You need to check out how Snow Leopard is built a little better, as it will not be 100% 64bit. It will be 'more' of a hybrid, but still not a full 64bit OS.
Actually, it will be a full 64-bit OS from kernel to user with legacy support for 32-bit carbon and cocoa. By your reasoning, XP 64 and Vista 64-bit aren't 100% either as both use the less-elegant WoW for 32-bit stuff and System32 for the 64-bit bits. This, in addition to the fact that Microsoft requires separate versions, rather than just shipping a product that covers both architectures. There's also that pesky problem that most programs for Windows are 32-bit apps, and many are very fragile on the 64-bit Windows platform. I'll stop now.
Because it DOESN'T matter in the Windows world. 32bit applications get performance benefits on the 64bit OS.
As it also happens with Snow Leopard, as the kernel goes full 64-bit (on Core 2 and newer machines). Current Leopard already gives the 64-bit benefits of increased registers and larger VM space to 64-bit programs, so there isn't a lot more to gain. These capabilities are not magically transferred to 32-bit apps under any OS, as the architecture is different, so there is no way that Vista-64 or Win7-64 will gift 32-bit apps with more than the 2GB of address space that they are currently allowed (vs. the full 4GB that OS X gives to 32-bit apps). Any speed increase comes from taking less time for register shuffles on 64-bit programs, giving more time to all processes.
Also if developers want to provide a full 64bit version, it is a simple recompile, you don't have to re-write the application like a lot of people (Adobe for example) find they have to do on OS X. This is why if you want a 64bit version of Adobe software, you need Vistax64, as the development APIs Apple sold Adobe never got moved to 64bit as promised.
Oh ho! What a canard. A simple re-compile was what has been hamstringing 64-bit graphics drivers and codecs for the last three years or so on XP-64 and Vista-64? Apple doesn't sell it's API's, it just publishes them. Adobe had 8 years of warning and opportunity to move it's cruft from Carbon to a cleaner, more modern API, and they decided to play chicken because Apple started eating their lunch with Final Cut Pro and Aperture. Being Carbon developers, they knew that it would not be an easy to clean up all of the legacy crap in the API without breaking backwards compatibility. Developing the iPhone was the final nail in Carbon's coffin.
All MS API sets(development platforms) move to 64bit, even old 16bit applications can be recompiled as 64bit applications. (You can't do this with System 9 applications, nor even the whole early 32bit transition APIs Apple provided.)
Yes, this is why Microsoft came up with thunking and the 16-bit \System folder and the 32-bit \System32 folder for Win32, which became the 32-bit \WoW and 64-bit \System32 folder. Brilliant. It isn't quite fair to drag System 9 applications into this mess. Do old Borland TurboPascal programs magically compile to 64-bit? Most System 9 programs were done as projects in Metrowerks Code Warrior with the PowerPlant framework, a system that was bought and buried by Motorola. Mac OS 9 was a cooperatively multi-tasked mix of PowerPC and 680x0 code, which used legacy system calls that needed Paschal syntax. OS X is basically a BSD Unix with no 680x0 code that uses GCC for compilation. Different architectures, compilers, system calls, hell the only things that are the same are the word "Mac" and "Apple". Up until the the Intel switch, you could run Mac OS 9 code, similar to WoW, without a re-compile. In most cases, the question is why would you want to.
Do you remember the Apple ads talking about the FIRST 64bit Personal Computer? How ironic that this many years later it still isn't even running a native 64bit OS, where Windows has been doing 64bit versions s
Re:I question the results. (Score:5, Informative)
But he could have benchmarked Vista and XP, then given an above/below rating for Windows 7.
And in fact, he HAS [zdnet.com] performed that test in the past and come to the conclusion that XP outperformed Vista.
The fact that his results are reversed this time must throw serious doubt on his credibility.
Re:I question the results. (Score:5, Informative)
The test you link to used SP2, while the new tests use SP3. XP SP2 and SP3 aren't the same thing. In fact, most benchmarks put Vista SP1 ahead of XP SP3 or at least within spitting distance of each other.
I'm not a big fan of Adrian, but he does hardware pretty seriously and lays out all his testing method well enough for you to duplicate it.
Re:I question the results. (Score:2, Informative)
Adrian does an enormous amount of pro-Ubuntu writing on his blog. Considering his closing paragraph contains
"And if youâ(TM)re put off by things such as activation and DRM, then Windows isnâ(TM)t the OS for you (good news is there are others to choose from)."
I don't think MS is going to be paying him any money for his posts.
Re:Still making 32 bit? (Score:3, Informative)
Unless you're playing tricks like storing other data into the upper bits of pointers, this shouldn't matter; from the application perspective addresses are 64-bit. With current hardware a bunch of bits is always zero, but allowing applications to use more memory should come transparently with newer hardware generations, with no recompilation necessary.
Re:Two reasons for this (Score:2, Informative)
Lots of people seem to think that Win32 API is a weakness of Windows when in fact it is its greatest strength. The stability of Win32 allows developers to target a huge range of operating systems with a single codebase (95, 98, ME, NT4, 2000, XP, Vista, 7). These days you only really need to support 2000 or perhaps XP and above, but being able to do that with a single binary API is a huge boon.
People are forever going on about how hard Win32 is to code against. Well, once you take the time to learn it it's not hard. Only if you aren't very talented would it be a problem. In reality most developers don't code directly against Win32, they code against a higher level wrapper (MFC, VCL, Qt, WinForms etc.) which makes it quite simple.
As a developer of commercial closed source software for Windows in a very small software shop I for one an hugely grateful for the stable and reliable development platform that MS has provided it. Without it we wouldn't have the successful business that we currently enjoy.
Re:I question the results. (Score:4, Informative)
"Right. Zero latency. Talk about lies. It establishes callbacks in the apps, writing into shared memory segments which are then mixed and delivered to the standard linux audio device."
That's pretty much all I want from my audio system. Just give me a precise control over my audio stack and then you can build anything on top of it.
XP does exactly this - there's a fast, efficient, hardware-friendly kernel streaming layer.
Vista on the other hand forces you to use inferior-quality stack because MS couldn't figure out how to do protected audio path with kernel filters.
"Yeah. Zero latency as long as you stay ahead of the playback. Just like pretty much every sound system since the days of the original Soundblaster Pro using DMA."
DirectShow is famous for its imprecise timing control due to KMixer ;) My previous employer made a lot of $$$$$$ by making time-correcting kernel streaming filters.
"Where's the signal processing layer in there? Oh, it's third party. Where's the channel synchronization? Can't find it."
Everything is third-party. JACK only gives you a microsecond-precise information about audio system. You can do the rest yourself.
"And shockingly enough it's all software. Where's that hardware acceleration you're so fond of?"
It's possible to have hardware filters in JACK. The problem is that hardware filters are not that useful for professional-type audio systems. Look at OpenAL/EAX for hardware acceleration of spatial sound and other goodies.
BTW, OpenAL Creative Drivers even work on Vista by bypassing all its audio stack.
And.... SURPRISE! Windows Vista uses 32-bit floats as internal audio sample format ( http://blogs.msdn.com/larryosterman/archive/2007/04/03/volume-in-windows-vista-part-1-what-is-volume.aspx [msdn.com] )!
"And what happens under load and the realtime scheduler can't quite keep up? Ah, I see, you get drop outs. What happens on Vista? Nothing, they hook into the scheduler to guarantee that their audio paths get time on the CPU."
Newsflash: if Vista scheduler can't quite keep up - you'll get sound drop-outs (I _do_ get them when I test my audio app on VMWare). There's no way around it. Realtime scheduler guarantees that your audio stack will get the highest priority, just like in Vista.
"It's not a matter of delaying individual streams. It's a matter of delaying individual channels from the same stream. So that your rear speakers sitting against the far wall behind you play just a bit earlier."
That doesn't matter. It's still not hard to do using kernel streaming.
I can distinctly remember that nice '3d-room' settings on my Creative Audigy 2 back in 2003. All in hardware.
About rebuilds (Score:5, Informative)
Do you still have to rebuild/reinstall modules for Linux for each version of the kernel?
In addition to the other /.ers' reports :
- openSUSE : No, you don't. .ko into the current modules collection.
if you install the drivers from an RPM (which is one single click on a web-page away, thanks to their 1-click-install feature) everything is taken care of by the package manager.
if you install the drivers from an ATI/NVIDIA installer or something more esoteric that you compiled your self, the openSUSE kernel upgrade will attempt (successfully in all my occurence) to import automatically the previous
- Debian stable : no you don't.
Everything including the kernel version, etc. stays the same across version updates, except for patched bugs. The previous modules keep working because the situation is exactly the same as before.
Atleast you don't have to reinstall every driver in Windows each time you've ran Windows update...
The fact that their whole OS stays exactly the same and doesn't improve a bit over the course of 5 years may have something to play in this situation.
Re:I question the results. (Score:3, Informative)
Unless you need the proprietary ATI or nVidia drivers, one reboot at the end of installation and it's done. And, if you do need to download those drivers, that's only one more reboot. Two at most, and you're done.
The only reason you'd ever need a reboot when installing Linux (apart from the obvous one to boot into your freshly installed system) would be if a new kernel had been released since your installation disk image had been issued and you have to upgrade. A kernel upgrade is the *only* case you have to restart a Linux (or pretty much any Unix nowadays for that matter) system (unless you managed to lock it up tight).
Video drivers are merely kernel modules (loaded dynamically, so no reboot) and a X11 server, thus requiring only a restart of the X server. So in layman terms, log out, restart your X display manager (or press Ctrl-Alt-Backspace, it will restart by itself), log back in and you're done.
Re:I question the results. (Score:3, Informative)
Optical media starts at the inside (i.e., closest to the hub).
Magnetic media starts at the outside.
Re:Still making 32 bit? (Score:3, Informative)
Since MS adopted the LLP64 [wikipedia.org] model, there really isn't a need to recompile 32 bit code to make them run on a 64bit OS. This model maintains maximum backwards compatibility but sacrifices it for forward compatibility. A 64 bit program would have to be rewritten for a 32 bit OS in this model. So companies would have to write and maintain two different source code trees for separate compiled versions.
Unix and Linux went with a LP64 model. Forward compatibility is stressed instead of backwards compatibility. In this model, companies would have to maintain two compiled versions but the source code would be the same but compiled differently.
Re:I question the results. (Score:2, Informative)
This is NOT a benchmark or any kind. It is a paid-for-MS-Win-7 advert. Seriously, no real performance tester would grade results as 1,2,3. WTF?
If you'd get that enormous chip off your shoulder for just a microsecond, you'd realize that the Win7 Beta EULA specifically forbids benchmarking. If he'd posted actual times or scores, MS can and would sue to get his site taken down. He did the only thing he could, which was posting 1st/2nd/3rd scores.
Instead of being so quick to scream "he's being paid by MS" (something, by the way, you have absolutely no basis for claiming), you'd appear less of an idiot if you bothered learning a bit about what you're frothing. I'll also remind you that Win7 is not available for purchase and won't be for another 6-12 months. Far from MS paying him to review Win7, they have a vested interest in not making people think they need to wait for Win7. Every person who sticks with XP while waiting for Win7 is a lost Vista sale.
Re:I question the results. (Score:2, Informative)