These drivers work great for me in Vista 64. Previous 174 leaks resulted in BSOD on boot (hacked 174.53) or lock-ups when I tried to alt-tab (174.70). I suspect there wasn't any sinister motives for witholding these 174s for mass consumption, they probably just needed to test them for older parts and work out some problems.
Here is the 60/40 split link.
Can you point out a quarterly result with the discrete GPUs only. The integrated GPUs run Vista so they are part of the driver errors.
Can you prove that all the reported driver errors come only from the discrete? In that case it's even worse for NV because of the high percentage of them.
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
and those people aren't running games, aren't overclocking, are probably not even running aero. The chance of a video crash when all you do is a 2d desktop is slim. I have seen a defective IGP before that crashed whenever a 3d game was ran. but could run the desktop indefinitely.
why does he have to back his statistic for you to retract something that is obvious FUD and based only on your assumption, opinion, and pro NVIDIA bias?
A year+ ago, Intel was clearly on top in terms of graphics in PCs. A year and a half ago x1950xtx was on top of the add-in world and AMD was doing well.
Right after the 8 series launch it is ridiculous to claim an assumption that most vista early adopters bought nvidia hardware because it supported DX10 as a reason for the elevated crash rate. to pass this off as fact is damaging to yourself and NVIDIA and to our forums.
how many people here overclock their graphics cards ... not most by a long shot.
now lets take that a step further ... how many people tried to overclock their graphics cards on Vista near launch when we were lucky to all of our drivers installed and have sound come out our speakers?
people pushing their hardware happens at a relatively low rate compared to the adoption of that hardware (with the possible exception being midrange intel CPUs just because they are way too easy to OC and just beg for it). combine that with an unstable new OS and people will steer clear of pushing their systems in order to increase their chance at stability.
and i'm sure you could "care less" what computers with integrated graphics do. especially if they crash at a lower rate than NVIDIA cards. Initially with vista and early drivers, gaming wasnt the only problem for me. I'd get graphics drivers resetting while surfing the web. intel, amd, and nvidia. you can't factor out 2d performance, as people experienced crashes here as well.
And how many people try to run games on an intel chipset? i bet a lot of people bought their "vista ready" computers and assumed it could run a game ... i bet a lot of people had a bad experience with their graphics, but it doesn't seem like that had windows crash on them.
dismissing intel and amd and saying that the increase in driver crash rates for nv are due to a ludicrously inflated market share is not something i'm going to let slide under the radar.
watch yourself.
Surely you knew this was gonna happen when you let him back? He's paid by nvidia to do exactly this. Will you be vetting all his posts? Sounds like a lot of work to me.
Good post, but it seems like you haven't followed his behavior since he returned here. This is nothing out of the ordinary for him.
While I don't necessarily agree with Rollo's approach, there's certainly some indicators out there that back his claim and directly refute some of the claims you make above. I linked to this earlier, but here's the Steam Survey with ~1.5 million unique samples. Its not perfect by any means, but its certainly solid evidence that would probably only be rivaled by Microsoft's own data. MS has released total number of errors by vendor, but without detailing % by vendor or even frequency per sample or even error type that doesn't do us much good.
You said yourself you experienced errors in Vista with parts from all 3 vendors, yet you never mentioned whether 1 vendor was more or less susceptible based on your experiences. As a reviewer you have access to more parts than most consumers so that feedback would've been valuable. There's plenty, including myself and Rollo, who have acknowledged TDR errors attributed to NV's drivers, but ultimately were fixed external the video card and its drivers. And again, that's before we even start talking about all of MS' hot fixes that largely undid the offending changes to Vista's video stack.
first, thanks, but we don't need that whole thing quoted in every post ...
the steam data is irrelevant. it was started nov 2007, which is well after most of the problems on all hardware came under control.
since you asked, i did personally have more problems with nvidia solutions than with any other. i feel confident in saying that our other reviewers had more problems with nvidia as well if the frequency of our internal emails on the subject are any indicator.
gary key and ryan smith would be the ones on the front lines with me with vista and i can ask them if you want a definitive answer.
yes, 3rd party apps and drivers did cause problems. yes ms hotfixes solved a lot of issues. but these things affected everyone. even if it might not have been nvidia's fault on a technicality, robustness in a system is still an important factor in driver design. if you are saying that these factors affected nvidia more heavily, you are also saying that nvidia's drivers are not as resilient as other companies' drivers, which is still an issue that should be taken into consideration.
the steam data is irrelevant. it was started nov 2007, which is well after most of the problems on all hardware came under control.
since you asked, i did personally have more problems with nvidia solutions than with any other. i feel confident in saying that our other reviewers had more problems with nvidia as well if the frequency of our internal emails on the subject are any indicator.
gary key and ryan smith would be the ones on the front lines with me with vista and i can ask them if you want a definitive answer.
yes, 3rd party apps and drivers did cause problems. yes ms hotfixes solved a lot of issues. but these things affected everyone. even if it might not have been nvidia's fault on a technicality, robustness in a system is still an important factor in driver design. if you are saying that these factors affected nvidia more heavily, you are also saying that nvidia's drivers are not as resilient as other companies' drivers, which is still an issue that should be taken into consideration.
How is it irrelevant? The only thing that would make it irrelevant is it emphasizes PCs used for gaming which you could argue limits the scope of potential errors to gaming applications. However, the data in the survey is completely relevant, when it started matters not as it is a snapshot of all hardware in that time frame purchased past to present. Sure there's going to be some migration as some PCs switch older parts for newer parts, but I can already tell you based on that Steam data the assumptions you made about the X19XX parts enjoying anything close to G80s success are simply UNTRUE based on:
1) Relative number of comparison 7800 and 7900 that remain compared to X19XX parts.
2) The % of 8800 parts in the survey. You can even exact the % of G80 as not to skew results with G92 by adding up the 320/640/768MB cards and subtracting that from the 8800 figures.
Coupled with a complete lack of competition in the DX10 arena until November (the 2900 and 38XX parts aren't even represented in most of the graphs as they're statistically insignificant) with your assertion that "most of the problems on all hardware came under control", its very clear that the high-end was in fact dominated by NV parts. And no, DX10 apps weren't a factor in crashes for Vista, but it was a significant check-box feature for those early adopters upgrading to NV and Vista going forward.
Thanks for the reply about which solution was more problematic. I wouldn't mind more feedback no, because again, you guys will have more experience than your typical end-user simply because you have access to hardware from all vendors. I do recall some of Ryan's writings on the topic in his Messy Transition articles and the memory allocation fix nearly eliminated all crashes in Vista. Now, if MS makes a change to their video stack that causes a video driver to crash, then turns around and hot fixes it and those crashes go away, who's ultimately at fault? I don't disagree with the possibility that NV's drivers are more susceptible to crashes, but again the fact that ATI and Intel both had ~10% and that major fixes came from MS to solve these crashes tells me you can't assume much at all from that 30% number. The main thing that bothers me is that there's those using this as an opportunity to take shots at NV's driver quality from that % without even owning NV hardware or ever using Vista.....
I didn't say that x1950xtx was anything near g80 level. I was pointing out that NVIDIA hadn't dominated the high end for about a year at the time Vista launched (g80 had dominated the high end for a year by the time the steam survey happened).
Additionally, if you think the under performance of 2900 XT and lack of a competitive high end part from AMD until THIS YEAR had nothing to do with NVIDIA adoption bewtween May and November of last year, then I honestly believe that you are mistaken (especially as the end of summer is a bit of a buying surge due to the "back to school" crowd). The steam survey says nothing about the state of the market ~9 months previous to its initiation.
When AMD failed to respond on the high end through out last year, NVIDIA gained a hell of a lot of mind share. I think its a very safe bet that the majority of Vista gamers and enthusiasts today own NVIDIA cards. But based on the information presented here there is no reasonable way you can stretch that back to Vista launch. None of this data suggests that everyone who has a G80 bought it at a time when Vista and drivers combined to make it a rocky experience.
I'm not trying to say that market share didn't play any role at all, as it could amplify any real problems that weren't isolated incidents. But Intel (with higher market share) was not similarly affected. My major point was to underline that market share is not a reasonable basis to write off the increased failure rate completely as some people seem keen on doing.
I agree that a lot of people can make a bigger deal out of the number than is necessary, and that you can't directly translate those numbers to an apples to apples per card rate of failure. But that doesn't mean it wasn't an issue at all; I have an easy time believing that NV's driver failure rates for early vista drivers before most of the hotfixes came out was higher than AMD or Intel's based on my own personal experience and that of other's.
Additionally, if you think the under performance of 2900 XT and lack of a competitive high end part from AMD until THIS YEAR had nothing to do with NVIDIA adoption bewtween May and November of last year, then I honestly believe that you are mistaken (especially as the end of summer is a bit of a buying surge due to the "back to school" crowd). The steam survey says nothing about the state of the market ~9 months previous to its initiation.
When AMD failed to respond on the high end through out last year, NVIDIA gained a hell of a lot of mind share. I think its a very safe bet that the majority of Vista gamers and enthusiasts today own NVIDIA cards. But based on the information presented here there is no reasonable way you can stretch that back to Vista launch. None of this data suggests that everyone who has a G80 bought it at a time when Vista and drivers combined to make it a rocky experience.
I'm not trying to say that market share didn't play any role at all, as it could amplify any real problems that weren't isolated incidents. But Intel (with higher market share) was not similarly affected. My major point was to underline that market share is not a reasonable basis to write off the increased failure rate completely as some people seem keen on doing.
I agree that a lot of people can make a bigger deal out of the number than is necessary, and that you can't directly translate those numbers to an apples to apples per card rate of failure. But that doesn't mean it wasn't an issue at all; I have an easy time believing that NV's driver failure rates for early vista drivers before most of the hotfixes came out was higher than AMD or Intel's based on my own personal experience and that of other's.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,002
- 126
Well if my XP experiences were anything to go on, nVidia's Vista majority driver crashes cannot be totally explained by other factors.
XP?s been around since 2001 so there's no excuses for hot-fixes, changing driver models or DX10 like are being thrown around for Vista.
At G80?s launch about one third of my entire gaming library had problems that significantly impacted my gameplay experience. What?s worse is that some of the games that appeared to be fine would hard-lock or slow down to a crawl when I actually sat down and tried to play them.
For about three months my 8800 GTS sat in my drawer because it was unusable while my 7900 GTX breezed through my compatibility demands.The 7900 GTX had what I consider gold standard drivers on XP so don't tell nVidia can't do the same with the 8xxx series.
The infamous alt-tab issue took months to fix and we still have Unreal 2 engine stuttering to this day on XP, a problem that?s been around since launch, in a TWIMTBP title no less. Not to mention that setting D3D pre-render to 0 or running 4xAA in either Vampire Bloodlines or Stalker will cause a BSOD without fail on XP to this day.
More current bugs here: http://forums.nvidia.com/index.php?showtopic=21028.
I?ve been maintaining this list since Nov 06. In Sep 07 I was told by nVidia it had been escalated to level 2 support, and I haven?t heard anything since and it?s now Mar 08. Not to mention Rollo has sent at least two emails to nVidia management and still nothing is being done.
?These are old games so don?t expect support? doesn?t fly. HL2 is as old as UT2004; is anyone really going seriously claim it?s okay for HL2 to stop working on the basis of ?progress??
Also ?it?s a unified model, so you should expect games to stop working? doesn?t cut it either. It?s kind of like Intel saying ?well, we?ve moved away from Net Burst so you should expect some of your applications to fail on Core processors?. Again, that doesn?t fly.
Even ignoring all of that, the fact remains we haven't had an official WHQL driver on XP for the 8xxx series since December 2007. Meanwhile ATi?s entire supported line-up got Cat 7.12, 8.1, 8.2, 8.3 with 8.4 coming next month. And they?ve been doing this since 2002.
No mater what spin someone puts on this, ATi?s driver support is simply superior to nVidia?s and has been for years.
To be fair, I don't like monthly driver releases. If I had it my way I'd ask for bi-monthly releases, and beta drivers work for me if I can't get whql on a schedule.
I disagree with your assessment of ATI drivers.
For years, one month's catalyst would break something that was fixed the previous month (they had (not sure what's up now) two driver teams working on versions that would leapfrog each other).
NVIDIA has almost always offered more customization options (until ATI finally listened and added clock adjustment and more quality sliders for things).
And when the Catalyst Control Center came out ... wow ... that was a piece of shit. Couldn't even try to start it til a good two minutes after windows started, and when you do try and run it, it would take forever to load. even navigating within the thing was slow and choppy. I don't see how anyone could ever have greenlit that thing in the state it was when it came out.
It's gotten much better over time, and I haven't noticed the same month to month issues with cats I have seen in the past.
In my opinion, NV had better drivers on XP (though I wasn't testing graphics drivers when XP launched), and ATI had better drivers on Vista until sometime in the second half of last year.
Currently I'd say they are on par with each other. Which I'm very happy with.
::EDIT:: oh yeah, and hooray for a move back toward unified drivers ... hopefully they'll be whql soon.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,002
- 126
I love them - it means rapid progress can be made and that's great for the consumer. In any case you?re perfectly free to only update whenever you feel like it but for those that need it the option?s there.
That certainly happened, but rarely. What I find with nVidia is when things are broken they generally take months if not years to fix, and this is particularly bad when the problem?s serious (e.g. the DEP issue on SP2 which took over six months to fix).
Heck, I was reporting the Alt-Tab issue back in 2004 on the 6800 Ultra and it?s only after the 8xxx series it was finally fixed three years later, after it had escalated to Vista no less.
ATi are generally much faster at fixing issues than nVidia, especially in legacy games which they don?t seem to forget as quickly as nVidia does.
Well since you?re talking about the control panels, both vendors suck and I use third party utilities. At least with ATi?s CCC you can access most 3D image quality settings right from the tray, unlike nVidia?s tray which doesn?t appear to have been updated since the GF 6xxx series.
nVidia?s control panel might not be as slow as ATi?s but it still sucks balls, especially for profile management where it?s particularly shocking. Half the time the settings don?t stick, take effect and/or show values that aren?t being applied.
I would disagree with this. After ATi got their act together with the Catalyst program back in ?02, I?ve found their drivers are more compatible and robust in general, especially in games and settings not actively benchmarked. nVidia appear to spend far too much time making drivers that produce pretty graphs for reviewers at the cost of everything else.
The 7900 GTX certainly had superb drivers and I?d love nVidia to be able to reach that standard again. My 9700 Pro and X800XL also had great drivers while my 6800 Ultra OTOH is one of the worst video cards I?ve ever had the displeasure of using.
As for the launch G80 drivers, they should?ve never been released in the state they were in and had the hardware not been so superior over ATi?s I would?ve sold it and gone back to them.
They were unified before and it still took months for a new driver. All that?ll happen now is 9xxx users will have to wait months for drivers like the rest of us.
Does anyone know if these 174.74 drivers resolve conflicts with old games for the 8800 GTS 512 (G92)? System Shock2 doesn't work on my 8800GTS 512, but it does work on my kids' 7600GT. Also there are similar issues with games like Thief 2, AVP2 and so on, although I haven't tried those. I currently have the 169.XX drivers installed. Can't use older (163.xx, etc) with these new cards.
i'm gonna have to disagree with this.
i run tests with almost every driver release (sometimes I'm unable like last year near the end of the year due to illness or whatever). but there was a period of year where every other driver broke the same thing ... for the life of me I can't remember what it was, but I do remember it was something that annoyed the crap out of me. i know that's gonna bug me ...
one of the things that had periodically returning issues was driver settings being applied. this was especially problematic with AA/AF. there was a while where you were required to reboot after changing AA/AF settings or crossfire settings in CCC to make sure they stuck. i can't tell you how many times i had to retest things because the settings didn't stick. The worst part about this on is that they would fix it and i'd stop rebooting. then they'd break it again and it'd screw all my tests again.
and then they fixed it for like 6 months ... and one day if freaking showed up AGAIN!!! ... it was just that one additional driver that had the problem, but it's been fixed for a long time at this point ... but i wouldn't put it passed them breaking it again.
I'm willing to admit that my perspective also has to do with the fact that the capability and ease of use of these drivers directly affect my ability to do my job efficiently. This might not make all my views of the overall subject matter relevant to the average gamer.
No, they're not anything to go on. Again, you're selectively taking anecdotal evidence and hearsay as the basis for your opinion that ATI's drivers are superior to NV's and this is another glaring example. Not only do you not use Vista, you never once complained about crashes in all your past rants about NV G80 drivers in the past. Considering how outspoken you have been about driver issues I *highly* doubt constant crashes and system hangs is a problem you would've tolerated as you've already conceded NV has the superior hardware numerous times.
We've already discussed your problems with NV's drivers in the past, but again we have to take a look at how relevant your experiences are when you haven't used an ATI part in 3-4 generations and 3-4 years. In previous threads you complained about NV legacy support for older games, like Thief and Thief 2, making completely baseless claims that ATI parts didn't have any problems. Yet it was revealed on these very forums that ATI's latest 2 and 3 series shared the very same problems. It was also noted that MS at some point deprecated DX support for palletized textures and that a 3rd party fix altered the game .exe and provided an alternative ddraw library to fix the broken calls. Those are application or API issues and while the 3rd party author acknowledges NV *could* fix them in their drivers, he's not entirely sure himself whether they're *obligated* to. As it seems all new parts have issues with that older title, are you now willing to concede it is in fact an application or API issue?
You also go on about ATI's flawless drivers and monthly driver releases. Monthly is nice, but how do you know they're not simply placebo? I mean honestly, would it satisfy you if drivers were released by Nvidia on a monthly basis and nothing was changed in them? Does that make any sense to do so? And before you go on about how ATI's drivers are improving each iteration, if they're so good to begin with, why release monthly drivers if there's nothing to improve upon? Its a double-edged sword and something I certainly wouldn't be trumping up without first-hand experience with them. As I saw in a reply by Derek, he has had quite a bit of experience with them and seems unimpressed.
But here's more evidence that you clearly don't know for any certainty that ATI drivers are superior to NV's:
ATI Hot Fix for Stuttering Issues in 3D Applications
Hot fixes must be all the rage now. As long as you meet your monthly schedule for placebo drivers, you're allowed to release intermitten hot fixes I suppose. But back to the addressed issue, where have we heard similar complaints about stuttering before? I have absolutely no clue what games suffered from stuttering as I don't run ATI hardware, and I'm sure you don't know either even though you've claimed ATI doesn't have any such problems. But it was certainly a big enough problem that they issued a hot fix to resolve it. Still so certain ATI never had any stuttering problems in any of the titles you droned on about over the last year?
But back to the original issue about the 30% crashes in Vista, we've already seen clear evidence from the Steam survey showing NV has a significant edge over ATI (2:1) in gaming hardware and a significant edge in DX10 capable hardware in Vista (70%). If you have more susceptible hardware out there running the applications most susceptible to crashes, common sense indicates your product will be more susceptible to crashes, and nothing else.
No Derek, this makes your perspective *ESPECIALLY* relevant, as you have direct access to hardware from all camps. I've mentioned this on some of review comments, that I wished reviewers wrote *MORE* about problems and conflicts they ran into during their testing but I know you guys handle a lot of that internally with the IHVs to resolve them before going to press. I know some of it is because you don't want to throw an IHV under the bus, but the reality of it is, the end-user is NOT going to get that red carpet treatment and tech support with open dialogue and custom/timely/updated drivers like you guys sometimes do.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter