Vista SP1 performance.

[Sharktooth]

I promised i would have been back on this topic when the final SP1 was out. So here we go. A preliminary test from CNET Labs and the proof of what i said months ago is here:
The long-awaited Service Pack 1 update for Windows Vista brings with it a number of significant bug fixes and other improvements. Microsoft says that with it you will also see a number of noticeable performance improvements as well. In our initial round of testing, we saw a mixed bag of results: we saw a few notable performance improvements under some conditions as well as significant performance degradations in others. For the most part, however, we saw little difference in performance between Vista and the Vista SP1 update.
Read more... (http://crave.cnet.com/8301-1_105-9871317-1.html)
So the question still is, XP or Vista?
I'd say: stick with XP (http://www.planetlowyat.com/blog/windows-vista-suck/)

[microchip8]

I'd say stick with Linux :P

[JohnnyMalaria]

But have *you* tested SP1? Do you always rely on some rag's "labs" to do the thinking for you? ;)

[Sharktooth]

i had all RCs of SP1 and tested them and reported there was nothing better about performance right here in the forum. when i said it didnt change anything you said to wait for the final. ppl said it couldnt be true (without any proof!)...
i have the final version now, tested it and it's the same.
now that ppl have the final version too, my impressions can be confirmed (read the article i linked in the first post).
SP1 is not increasing any performance in normal conditions (except some file copying issue that is now even worse when copying stuff from removable devices).
Vista is still the same POC as ever... even with SP1.

[CruNcher]

Sharktooth this was a base Vista installation with all Services and Aero running?
i wonder how Vista performance is with a normal base NT shell (Windows 2000,XP) and maybe all the new loging mechanisms off so practicly only the new enahanced Kernel and Memory Management that Mark Rusinovich describes in his talks :)
I find it a little unfair how Vista is being compared vs XP from the most they should compare both against each other with the same services running and the same shell so basicly with the same overhead for both. Especialy Priority processing should be much better with Vistas Kernel then XPs (no system slowdowns when hardware fails that is on the same IRQ then other Devices for example) <- this can also easy reproduced with a defective cd that canot read a file correctly from it so the whole system beginns to slowdown (no IO acces to other devices possible or getting very slow Vistas kernel (priority system) should be smarter in such cases, if i followed the development correctly).

[Dr.Khron]

I think Mossberg did an excellent job of testing the new perfromance claims for Vista SP1:

I tested three of the performance improvements Microsoft claims for SP1. The first involved speeding up the copying of hefty folders containing large numbers of files. On both machines, copying a folder containing over 700 files totaling almost 700 megabytes took less than half as long with SP1 as it had with the original Vista.

I also tested how long it took both machines to awaken from a hibernation or sleep state and be ready for work. For these tests, I began with each machine running Microsoft Word, Microsoft Outlook and the Firefox Web browser, then I forced them into sleep and hibernation mode.

By my definition, "ready for work" means that Vista's circular delay indicator has gone away, the software that loads at start-up has finished launching and the computer has fully reconnected to its wired or wireless network. On both of my test machines, SP1 improved the recovery time from sleep or hibernation, shaving one to 10 seconds from the procedures.

Microsoft doesn't claim SP1 will improve the speed of cold starts and reboots under Vista, but I tested these anyway. To my horror, I found that SP1 actually made rebooting -- already slower than on comparable Windows XP computers or Macintoshes -- even slower.

Microsoft explained that this was due to the fact that installing SP1 erases certain data used by Vista to speed up program launching. It takes the system a few days to build this data back up, the company says. Until then, it says, overall performance, including reboots, can be slower under SP1 than under original Vista.

Microsoft provided me with a method that would rebuild this program-launching data more quickly, at least for the common programs I was using in my tests. Once I followed that method, rebooting time returned to its former state -- still too slow for my taste, but at least not worse.

Granted this guy is far more "mainstream" then "hardcore", but at least he never kisses MS's ass like most of the mainstream press do.


EDIT:
Full link: http://online.wsj.com/article/SB120295315055367067.html
There is also a video clip.

[Sharktooth]

At the time i did the tests i disabled everything could be disabled except superfetch.
I streamlined the SP1 it into the Vista iso and burned the DVD.
Then i compared the fresh installations performance with sintetic tests as well as common benchmarks and common operations (softwares installations, softwares launch times, file operations, etc).
Vista RTM, updated Vista RTM (patched except SP1) and Vista SP1 have comparable results.
I noticed Vista SP1 has a slightly higher memory "footprint" than the other two.
Also patched Vista RTM has a slightly higher memory footprint (not as high than SP1 though) than Vista RTM as well.
Those tests were made on Vista Ultimate.

[CruNcher]

I just can't belive that Microsoft drifts so far away from Performance in what they achived with 2000 and XP in system stability (sure performance also lacked but that's normal for more stability see the NX execution thing in software with XP SP2) compared to 98 and Me it's somehow unbeliveable for someone who watched the whole development and the improvements that where done to this OS over the years. Ok the Protected Media Path and all the Stability Loging + (how much Services are actually new introduced?) costs alot of new IO accesses and (Aero Memory) for sure but it shouldn't make a system with a enhanced Kernel that much slower then it's predecesor i just can't belive it, even if i hear it almost daily :). And especialy it shouldn't be slower if you count the fact that alot of the Shell is now GPU accellerated so the CPU gets even lower utilized now it should be faster not slower, someone should realy compare the final RTM of Server 2008 vs XP :D

Someone should log especialy all IO accessing that takes place in Vista compared to XP (IO access is still something that can slow down subjective work speed and some could even think memory is the problem then)

[Ryokurin]

not calling out your experiences, but you did do a Reverse Integration instead of a slipstream (http://www.winbeta.org/comments.php?catid=1&id=9536) right? because at the moent slipstreams are broken.

[CruNcher]

I didn't test it so i can say nothing about it im since the whistler betas on XP and im happy with it, but i keept a close eye on the development of blackcomb and longhorn (especialy kernel improvements), and what i heared (before) and now after the release of the final product named Vista by the users shocks me to death (especialy as MS developer wise it was allways held up so high) i thought it would be a little slower but not that heavy slower like people reported in beta and final, it sounds crazy like a complete rewind of all the XP performance improvements.
And sometimes i feel like 8 year old child playing with such a laptop with Vista installed but only Graphicly usability wise i don't see such big inovative improvements @ all not the likes that Apple for example integrates all the time into MacOSx (where is Microsofts integrated Virtual Desktop Combined Workflow like Spaces??? Flip3D good joke) even Linux with Metisse has it and Microsoft lacks something such inovative you could just cry, but yeah some like to watch their animated Ultimate background instead WTF???.
So it doesn't really suprise me that Companies don't change to Vista nothing improved really in the Workflow if you call the PictureGallery or how it's called a improvement @ all with it's panorama function (jeez please), only little improved thing in Vista maybe is the Speech recognition (human interfaces) usability wise but even that is nothing really inovative ( Dragon Naturallyspeaking for XP and you set) (the video that's floating arround is like the Windows98 Sattelite Video plugin USB device and Plug & Play bluescreen, ok not as hard as it's no stability problem like @ that time but as funny, because it was also a official presentation) :D

[Sharktooth]

not calling out your experiences, but you did do a Reverse Integration instead of a slipstream (http://www.winbeta.org/comments.php?catid=1&id=9536) right? because at the moent slipstreams are broken.
yes. exactly. i found that link in the vlite website and followed the instructions since i failed using vlite to integrate the SP1 on the Vista image... so...

[Morbo]

SP3 for XP is a step up though.....;)

[shon3i]

SP3 for XP is a step up though.....;)
SP3 for XP is been only rollup package, no any performance changed since SP2 IMHO. I now use RC2 3300

[Sharktooth]

A tip for "restoring" network performance as in WinXP...
Once again, Microsoft thinks it can solve all of your problems by being smarter than you are. Unfortunately, some of the "automatic tuning" features of Vista can yield in lower performance.

One of Vista's features tries to change the network's receive window on -the-fly to adapt to changing network conditions. If you've been experiencing lower network throughput when transferring large files or experience weird network disconnects, then you can try disabling Vista's auto-tuning network feature.
Read more... (http://www.ocmodshop.com/ocmodshop.aspx?a=900)

[jmnk]

A tip for "restoring" network performance as in WinXP...
Read more... (http://www.ocmodshop.com/ocmodshop.aspx?a=900)
Hmm, interesting. Basically the tone of the article suggests that Vista is at fault here because it tries to optimize network connections using standardized feature that has been around as RFC since 1992 - because 'some' routers may still not support it? Even linux kernel since like 2.8.17 has this feature turned on - and I have not seen you advising people to turn it off on Linux??

[dloneranger]

I've being playing about with the sp1 release today, and on the whole I'd give it a thumbs up

I'll take the slightly degraded performance of usb external drives for being able to change the network throttling while playing any audio, that was killing me until today
Finally, back to xp performance, nice to have 50Mb/s back again

[Sharktooth]

Hmm, interesting. Basically the tone of the article suggests that Vista is at fault here because it tries to optimize network connections using standardized feature that has been around as RFC since 1992 - because 'some' routers may still not support it? Even linux kernel since like 2.8.17 has this feature turned on - and I have not seen you advising people to turn it off on Linux??
on linux there's no need to turn it off coz it works.... while the Vista implemetation seems buggy.

[burfadel]

I have to say Vista SP1, and even before SP1, is far more stable on my computer than XP ever was, and my friends have experienced the same thing. I have to push the fact than I'm using Vista x64, I think 32 bit vista is an absolute waste of time and would recommend sticking to XP for 32 bit mode. For x64, windows XP x64 which was based on server 2003 x64 (and used the same updates) was in itself much better than XP, and XP x64 and Vista x64 seem to be about on par for most things performance wise. Admittedly some things are slightly slower on Vista, but having more RAM gets around that - that is, the performance benefit of extra RAM is greater for Vista x64 than for Xp x64. I saw someone running 8GB of ram (which you can with x64 and a bios that supports remapping), that dual booted xp 32 and vista (64 bit of course) and the performance gap was very close. Windows Xp x64 is much faster transferring files than XP, it would be very nice if that performance was carried through to vista!

The file operations were still slower on Vista, but other things were faster! I'd recommend no less than 2gb, and it should be noted many of the comparisons only use 1gb of Ram which naturally favours XP.

Vista for me is MUCH more stable than XP, and any display driver issues with nvidia have been resolved with newer non WHQL drivers, currently using 169.61. Many of the comparative tests use up to date drivers for XP and in the box drivers for Vista, again that is a very false argument.

I'm not saying Vista is perfect, just that in terms of stability with the advent of sp1 it is easily equal to xp x64 and superior to 32 bit XP. Again vista 32 sucks and should be avoided! - which is hard since its the default version for OEM's, and upgrade packs (with upgrade packs you have to specially order the x64 disks at a shipping charge. why not just include the <expletive> disk to start with)!