I thought it was "snugglebunnies"?The Baron said:
Driver stability - Carmack still gives edge to NVIDIA
- Thread starter Reverend
- Start date
No, Dig, you're my snugglebunny (or at least you are until you occasionally really piss me off and then I yell at you and then everything's okay). Stealth is my snugglebuddy. Get it straight.digitalwanderer said:
(Hanners, you're... uhhh... something.)StealthHawk
Regular
The Baron said:Hehe. Seriously, sometimes I think Stealth is me and I am Stealth, and we are simply polar opposites in our perspective. It's funny.cthellis42 said:The Baron said:
I think there are other forums for that.
It's all good discussion though
I'm not disagreeing with you. Although I haven't seen many problems with my ATI card(the few that I have were software problems, not driver problems), there are many who do. NVIDIA's drivers are still generally more stable from an enduser's point of view. Although lately they seem to be a little buggier while ATI has improved.
Some bugs, of course, seem to be ATI's fault and no one else's. Like the system corruption when choosing to optimize for cache in winXP. How something like that crept in(and is still unfixed?) is kinda sad.
StealthHawk said:
Bah! I expected more out of you by now Stealth.
The system corruption bug when you select Optimize for System Cache in XP can be caused by just about anything and it has various degrees of severity too. I once had a system work wonderfully for a month or two...put Norton Anti-Virus on and changed it to System Cache while tweaking my system...well that was the last I ever saw of that install. I couldn't even change it back in safe mode as it kept complaining about a delayed write failure.Granted I just mentioned one specific program which is known to cause this problem from time to time (it really is a random thing) but it can happen to almost any system due to any piece of software. It is one of those XP oddities. Since there is no real way to predict its occurence or what piece of software will cause it I have since stopped using that tweak. It isn't worth the risk of it possibly killing a system at a critical time. For instance I would be in deep you know what if it took out any of my systems this week as they are holding a major lab writeup due Friday, a major design project due then too and various other school projects due in the next week or two that have taken countless hours to produce. (Lab is looking at roughly 100 hours for completion
No sleep for me)Enough of me rambling...point is it isn't an ATI specific issue and I don't think anyone has even figured out exactly what causes it yet (other than not optimizing for System Cache).
Regarding the System Cache bug with ATI drivers: well, they have a little note about it in the readme that pops up. so, really, why don't they just add a little program that checks the registry, and if it sees that System Cache is selected, it prompts you with a box that says, "HOLY SHIT DUDE! YOU NEED TO TURN OFF OPTIMIZE FOR SYSTEM CACHE, OR YOUR COMPUTER WILL EXPLODE!" I mean... really. Don't rely on the users to actually read documentation. Users are stupid. They won't have any idea that something applies to them until A. it happens to them or B. the program notifies them that it will happen to them. So... yeah. I'm done.
StealthHawk
Regular
The Baron said:
Exactly. The problem obviously does happen as ATI has documentation and warnings not to use the setting! It obviously does happen frequently enough to warrant such warnings in the first place. I don't think ATI would list it as a problem if there were only a few isolated incidents.
re Razor04: I am not aware of any other IHVs drivers having the system cache problem. At least, no other vendors publically state that such a problem can occur to my knowledge, and that a particular winXP setting should be avoided. Therefore I can only fairly assume that it is an ATI problem, since I have not heard of other combinations of drivers and software causing it.
If you can show proof that NVIDIA, Matrox, or what have you drivers can also lead to system corruption then I might be inclined to believe that some other software or even winXP itself was to blame.
edit: I checked the Catalyst driver bug list at R3D and apparently this is an OS bug. Fair enough.
Wow, I've always wondered why people say nvidia has such good drivers....
I had a geforce 3, and it was always crashing, all sorts of instability and graphical errors.
Now I have a radeon 9700 pro, and it never crashes, almost never has graphical errors, only has minor bugs(I had to use a 3rd party utility to get tv out working on my geforce 3, and always had troubles installing the drivers), such as not matching resolution correctly to my screen(I have an old coporate monitor, it supports up to like 1800 by something at 100 hz, but if I force higher refresh rates than 60 hz, I get rather large black borders on the screen, I guess it's probably using the monitor's max resolution as the standard res, but the provided controls to adjust the image don't do much to fix it.
I had a geforce 3, and it was always crashing, all sorts of instability and graphical errors.
Now I have a radeon 9700 pro, and it never crashes, almost never has graphical errors, only has minor bugs(I had to use a 3rd party utility to get tv out working on my geforce 3, and always had troubles installing the drivers), such as not matching resolution correctly to my screen(I have an old coporate monitor, it supports up to like 1800 by something at 100 hz, but if I force higher refresh rates than 60 hz, I get rather large black borders on the screen, I guess it's probably using the monitor's max resolution as the standard res, but the provided controls to adjust the image don't do much to fix it.
The Baron said:
Wow...I don't know what in your hardware configuration didn't agree with your 9600P, but I've been running the Cats since last Sept., beginning with a 9700P and then a 9800P back in May, across two different motherboards and chipsets (KT333 & nf2) , and at least three different cpus (Athlon Palomino, T-Bred, and Barton), with WinXP and service packs and all subsequent patches, and during that time have run--oh, maybe 35-45 3d games (when I first bought the 9700P I pulled out much of my software library and did some serious compatability testing and posted the results last year in a R3d forum--didn't find one that refused to run, or that crashed due to obvious driver bugs), and as memory serves I can only recall one problem I had with the drivers in an early Cat release, which was fixed next release. My experience with the Cats has been nothing like yours at all--which leads me to believe you had some other problem, possibly relating to your general hardware-software configuration, maybe, which for some reason doesn't affect your 5700 (the cards require different hardware/software resources, so it could be many things.)
I honestly can't recall the last time I crashed due to a problem I could pin on the Cats (and it's very rare that I crash at all), so I'd have to emphatically disagree with your assessment--the Cats have been exceptionally stable for me, and I've had no cause in the past year to even consider going back to nVidia, after swapping out a Ti4600 for a 9700P last year. That's why I bought a 9800P in May--because my experience with the 9700P had been so positive (I am definitely not a glutton for punishment and certainly wouldn't have bought R350 if I'd had trouble with R300.) More than a year of stable, almost trouble-free operation, just as stable and trouble-free as I ever knew with nVidia hardware and drivers, is exactly what I've had. Your experience certainly does not parallel mine, is all I can say.
I don't crash often enough to be able to recall them, so I really can't address your comments on the "VPU recovery" feature--except to say I wasn't crashing before I installed the drivers which installed the VPU Recovery feature, either. So, it's not something I give any thought to, really. I do find it kind of amusing, though, that people might assume that because an IHV puts GUI control of this feature in the CPanel and places its activation under user control, that this is some kind of "proof" that the drivers are unstable and the company "knows it." Considering the fact that the Forcenators do several things that we know of despite the wishes of the end user as expressed through the Forcenator Cpanel, I think it's very possible that nVidia has something similar in its own driver code--the difference being, of course, that you don't know about it and are not consulted as to its use or activation.
_________________________
And reading further into the thread I see the ever-lasting topic of "system cache" is resurrected once again. I run desktop systems on a network at home, and those machines see a lot of 3d gaming, and are not used as dedicated servers, which is why I guess I've never felt compelled to setup XP for anything except standard "program cache" memory management operation. Since I don't run dedicated servers at home, it never seemed a good idea to me to instruct XP to run in "system cache" mode, a mode which cannibalizes ram on those machines from the programs and games which I run which need the ram for themselves and turns it over to system and file operations which function perfectly for me at their default "program cache" settings with regard to memory management allocations. Ram-starved programs and "stability" just don't seem to mix, IMO. Dedicated servers often have a couple of gigs or more of ram, which is really the memory paradigm "system cache" needs in order to throw a lot of available system ram at system operations without interfering with the ram-hungry programs that may be running at the same time, it appears to me. I've got a gig of ram in each of my home machines, which is a decent amount for a desktop, but not nearly enough for a dedicated server that's going to have work fairly hard as a server, IMO.
Early on after WinXP began shipping, I ran into more than one so-called "tweak" site which was telling people that "Changing XP from program cache to system cache is a performance tweak," while at the same time never even bothering to explain to their readers what the settings mean, and what the difference is between them. Heh...
I guess it was a case of "look at the pretty button, press the pretty button, and now the system is changed, so it has to be better than it was," I suppose. I mean, how someone might promote something as a general "performance tweak" on a web site without understanding it well enough to tell his audience *why* it's a "performance tweak," instead of something else, is beyond me. Often though, that's the case with a lot of "tweak" advice, which basically operates under the assumption that all default settings in an OS as it ships are wrong, and that the company shipping the OS deliberately wants to hobble its performance in the hands of its customers simply to be disagreeable and rob them of "the performance they deserve"... (Anecdote: I have a friend who followed some XP "tweak advice" from one of these sites last year concerning all of the "unnecessary" services running and how his life would be so much better if he turned off most of them, and the site provided a list of the services the author said were "safe" to turn off, because the author had no idea of what they were for and concluded they were "meaningless" and just "burned up system ram needlessly", etc. Thinking that the person offering the advice was a "pro" my friend dutifully followed the instructions, and upon the next boot XP took about 6 minutes to get to the desktop, his network connection was gone and of course also his broadband, and his system ran so slow it was barely navigable [actually the guy is my neighbor.] He called me and I dropped by next evening and fixed him up and admonished him "sternly" not to believe everything he's sees printed on the Internet...
Afterwards, out of curiosity I visited that "tweak site" and was pretty amused in reading some of the descriptions the guy had written up. I guess he figured that everybody's machine is a clone of his own and that what worked for him was a "tweak" that would work for everybody, even though he obviously didn't understand turning off services beyond the idea of "freeing up ram." There are certain services which I disable, but only because I understand them fully enough to know that in my case I don't need them or want them running. Anyway, I'm all for legitimate "tweaks" that actually do something useful, and there are definitely such tweaks in circulation, and I use some of them myself. But unfortunately there is as much bad advice dispensed about "tweaking" as good advice, I think.) A lot of people in those days were indoctrinated with the idea that invoking the system cache memory management mode for XP was indeed a "performance tweak" for normal desktop environments, and one of such importance that they could not and would not give it up. I've read at least a couple of dozen posts from various individuals who would endlessly reinstall the OS, driver sets for devices of all types, and some, even, who swapped out perfectly good 3d cards which didn't work properly under system cache memory management mode (but work perfectly under the default program cache memory management mode of the OS), simply to hang on to their fixation with the "system cache" setting being a "performance tweak" of such magnitude that it affects their entire desktop computing and 3d-gaming experience...
(Which I've seen no evidence to support.)Yea, setting this inappropriately may *definitely* affect one's desktop experience, although maybe not in the desired way. As Razor04 points out, there are certain applications which simply don't like the OS trampling on their memory demands for the sake of handing off ram to the system, and don't do well in desktop machines of limited system ram (compared to the amount of ram the system cache mode expects, the amounts typically found in dedicated servers.) To my way of thinking, ATi's drivers are no different than those applications in that they are written to expect certain things relative to the memory management of XP, things which *change* when the system-cache setting is invoked.
I think ATi has been more than clear over the last year in explaining that it does not tune its drivers for its 3d cards to run in the XP system cache mode, and they've done so rather unapologetically. I think they are entirely justified in this position, as it seems a bit ludicrous to me that people and companies running dedicated servers would set them up with $300-$500 3d cards so that they could run 3d games at the same time. Heck, many a dedicated server motherboard is missing an AGP slot completely, and runs cheap integrated graphics of a type designed to interface the server without regard for an ability to do much else, including playing 3d games. It's for this kind of dedicated server environment that I think the system cache memory mode under XP is intended to serve. And as ATi has stated that they are aware of it and aren't going to change it, it is also not a "bug" as it is often inaccurately labeled. Obviously, it's a deliberate design decision on the part of ATi which specifically concerns its 3d products and reference designs and drivers. I don't see anything wrong with ATi's thinking here, but I can see a problem with people fixated on setting the system cache memory management mode for desktop machines built to, among other things, run 3d games.
I'm sure, however, that the ability to run their desktops in system cache mode is more important to some people than the type of 3d card they run, and for those people I would definitely recommend against ATi as the company has made it clear their drivers are specifically set up for the default XP program cache mode, and it's that memory-management environment the Catalysts expect. So, if the system cache setting is the most important thing to you, then use something else other than ATi, even if your 3d performance and IQ suffer as a result, since obviously that's merely a secondary concern anyway. For my machines at home which are definitely not dedicated servers, but desktops I use for 3d gaming among other things, I've never been tempted to use the system-cache setting, whether running nVidia or ATi 3d products, and haven't done so. I just don't see much point, or fun, in trying to hammer square pegs into round holes, most especially on the basis of "performance tweaks" which are based on erroneous assumptions in the first place (if, that is, there's actually any assumption being made aside from the idea that default settings "always degrade performance," or something similar.)
WaltC, I agree 100%, the people who do have problems shouldn't squarely blame it on the video card.
I have had 0 bugs and 0 problems with my R300. Absolutely no bugs.
Another thing, I am able to run the 'system cache' tweak fine on my system. I have already ran it twice without problems that others see. Though you are right WaltC, there are no benefits from doing so. I'd rather leave all my ram for the program to hog.
I have had 0 bugs and 0 problems with my R300. Absolutely no bugs.
Another thing, I am able to run the 'system cache' tweak fine on my system. I have already ran it twice without problems that others see. Though you are right WaltC, there are no benefits from doing so. I'd rather leave all my ram for the program to hog.
Althornin said:
I dont think you can boost your gaming performance by windows tweaks, but you can speed up their bootup times. My XP is usable as soon as desktop shows except network connections that require additional 10 seconds.
Untweaked XP boots, shows desktop and lets you move mouse freely but when ever you try to run something via start it shows busy icon.
That is the main difference between tweaking and nontweaking, although you really have to know what you are doing.
Zvekan
Going OT, but still roughly on topic considering the original subject...
Being a FX5900R user, and before that a Ti4200 user, well...
1.5 years ago, I loved NVIDIA's drivers.
And now, for over a year already, I hate them. The word "crap" describes the changes they've done perfectly. They're still good drivers, don't get me wrong, but their quality has obviously been going down.
Yes, they're stable. But some of the bugs in them are just plain ridiculous.
They needed 6-9 months to fix MSAA on both my Ti4200 and FX5900R, and that in several games, including Morrowind; not only that, but the few beta releases that did fix it often had other major problems ( like one which fixed MSAA in Morrowind, but which made PS water awful looking ). To their benefit, it's 100% fixed now Too bad I don't play Morrowind anymore, and I did back then
Also, the FX5900's scrolling noise is okay most of the time, but in certain application and games, it's ridiculously loud and annoying. I've yet to even see NVIDIA acknowledge that problem by something else than "Really? Hmm, strange. Could you send us some more detailed info please?" in an old IRC chat... *sigh* No, it's not all FX5900s, and it's most likely related to some minor hardware conflicts - but it's clearly either a driver bug, or worse, *gasp*, an hardware bug only applicable on certain chips for extremely obscure reasons.
Uttar
P.S.: Just to make myself clearer, do not believe I hate NV's drivers and it'd be a serious reason to switch to ATI before my normal upgrade cycle. NVIDIA's drivers are still okay IMO, but way worse than they were 1.5 years ago, before the notorious Det40 release, which brung only bugs and the 3DMark01 Nature cheats...
But what this does mean, however, is that I'll never consider ATI's not-as-good-as-NV's-driver-reputation-but-improving disadvantage again; I'm sure ATI's drivers must be as good, if not better, than NV's nowadays. Which means I can just compare performance next time around; a refreshing change indeed
Being a FX5900R user, and before that a Ti4200 user, well...
1.5 years ago, I loved NVIDIA's drivers.
And now, for over a year already, I hate them. The word "crap" describes the changes they've done perfectly. They're still good drivers, don't get me wrong, but their quality has obviously been going down.
Yes, they're stable. But some of the bugs in them are just plain ridiculous.
They needed 6-9 months to fix MSAA on both my Ti4200 and FX5900R, and that in several games, including Morrowind; not only that, but the few beta releases that did fix it often had other major problems ( like one which fixed MSAA in Morrowind, but which made PS water awful looking ). To their benefit, it's 100% fixed now
Too bad I don't play Morrowind anymore, and I did back then Also, the FX5900's scrolling noise is okay most of the time, but in certain application and games, it's ridiculously loud and annoying. I've yet to even see NVIDIA acknowledge that problem by something else than "Really? Hmm, strange. Could you send us some more detailed info please?" in an old IRC chat... *sigh* No, it's not all FX5900s, and it's most likely related to some minor hardware conflicts - but it's clearly either a driver bug, or worse, *gasp*, an hardware bug only applicable on certain chips for extremely obscure reasons.
Uttar
P.S.: Just to make myself clearer, do not believe I hate NV's drivers and it'd be a serious reason to switch to ATI before my normal upgrade cycle. NVIDIA's drivers are still okay IMO, but way worse than they were 1.5 years ago, before the notorious Det40 release, which brung only bugs and the 3DMark01 Nature cheats...
But what this does mean, however, is that I'll never consider ATI's not-as-good-as-NV's-driver-reputation-but-improving disadvantage again; I'm sure ATI's drivers must be as good, if not better, than NV's nowadays. Which means I can just compare performance next time around; a refreshing change indeed
Well I still hate nVidia, clear and simple. 8)PaulS said:
PaulS said:
Sorry, I know I wasn't clear, should have reread that post :?
I meant I hate the changes NVIDIA has done to its drivers ( control panel is not particularly good, just okay, and the overall quality has decreased overall ).
So I hate the changes they made, but they were so good in the beggining I still find they "okay", maybe even "kinda good". Just certainly not better than ATI's drivers ( from what I'm hearing, could be 100% wrong there; ATI's could be way better or way worse, I never had first hand experience, but for a Rage years ago ).
Uttar
I have absolutely no experience with ATi drivers, so this is really just an impression I've gotten from the various forums around, but it seems to me that the ATi drivers are finicky and very susceptible to hardware conflicts. But if you've got a system which avoids the particular troublesome hardware combinations, they're great - and probably better than their nVidia counterparts.
The monthly driver release is definitely a plus in my mind, along with the hot fixes - Having to wait months for a fix for your favourite game/app just isn't so hot, ya know? And all this makes it kinda difficult for me, moving forward. If I was buying a GPU now it would be a no brainer, but what about 6 months down the line? If the NV40/R420 were about equal in terms of performance, who would I go with? Stick with what I know (nVidia), but risk waiting months between driver releases (and the fixes), or go with ATi, hoping I don't run into compatibility problems, but safe in the knowledge that the next driver is only a few weeks away?
Again, this is all an impression. I may well be completely wrong.
The monthly driver release is definitely a plus in my mind, along with the hot fixes - Having to wait months for a fix for your favourite game/app just isn't so hot, ya know? And all this makes it kinda difficult for me, moving forward. If I was buying a GPU now it would be a no brainer, but what about 6 months down the line? If the NV40/R420 were about equal in terms of performance, who would I go with? Stick with what I know (nVidia), but risk waiting months between driver releases (and the fixes), or go with ATi, hoping I don't run into compatibility problems, but safe in the knowledge that the next driver is only a few weeks away?
Again, this is all an impression. I may well be completely wrong.
Similar threads
