Nvidia drivers v29.42 released - Aniso performance

[Guest PaulL01]

Elrond,>Ok, instead of responding line by line, let me submit a set >of proposed facts instead. Maybe we'll get to the finish >line easier this way. Either agree or disagree: Finish line? Then wot? :-lol>>1) The v29 series fixes OpenGL Aniso performance on the GF4. > Still has room for improvement of course. Agree, but this was fixed a while back at 28.92 (two months ago)>2) In OpenGL apps and games, Aniso performance is also >improved with the v29 series drivers on the GF3 and GF2 >platforms. As compared to 27.xx-28.xx drivers running GF3/GF4 agree, compared to GF3's running 22-23.xx completely disagree, not bad just not better. As for the GF2 I can't see how or why we are discussing this card as it doesn

[Guest PaulL01]

>Elrond it just occured to me that once again the refresh >rate is not working on these with win2k.This is seems to be >a new trend in nvidia drivers which is quite annoying to say >they least so I have gone back down to the 23.11.Iam hoping >that they will fix it :/ other than that I do think in some >areas the 23.11 are a better performance driver but the >29.42 give a better picture. Refreshrate problem is with win2k/winXP and MS just announced that it will be fixed in a SP release or a DX upgrade, but who knows how long that will take! :-rollPaul

[Guest]

Hi Richard,Paul has it right: its not an Nvidia driver problem, its a Microsoft problem. They changed the way refresh rates are handled in 2K and XP - which is to say - they broke it. They have said it'll be fixed in the next SP.In the mean time, you can fix the refresh rates for any Nvidia driver (including the latest) on 2K/XP using the NVRefreshTool. I wouldn't fly without it:http://www.nvrt.netGood luck,http://members.rogers.com/eelvish/elrondlogo.gifhttp://members.rogers.com/eelvish/flyurl.gif

[Guest]

Hey Paul,"Finish line? Then wot?"One of us gets the Cadillac that Tom has put up for us... :-lol">1) The v29 series fixes OpenGL Aniso performance on the GF4. > Still has room for improvement of course. Agree, but this was fixed a while back at 28.92 (two months ago)"Yes of course. But we're talking released, stable drivers here for recommendation to users. Or so I assumed this thread was all about."Are we on the same page as far as AF testing? That is: enabling 32/64 tap AF as that is the problem. Or am I wrong on the GF2GTS?"No, we're on different pages. As I've said previously, all code optimization that is good is good code optimization for all boards. 8, 16, 32 or 64 tap: no matter. As the code gets optimized for Aniso, improvements appear in the lower end cards as well since the entire code base is unified - just not as dramatic as the higher end cards. I'm guessing you have to be a developer to get this idea.Argggg Matey! No finish line in sight! :-) Ah well, I spose theres nothing wrong with two differing opinions. Specially when I know mine are the correct ones :-lol :-lol!Take care,http://members.rogers.com/eelvish/elrondlogo.gifhttp://members.rogers.com/eelvish/flyurl.gif

[Guest PaulL01]

Elrond,>"Are we on the same page as far as AF testing? That is: >enabling 32/64 tap AF as that is the problem. Or am I wrong >on the GF2GTS?" >>No, we're on different pages. As I've said previously, all >code optimization that is good is good code optimization for >all boards. 8, 16, 32 or 64 tap: no matter. As the code >gets optimized for Aniso, improvements appear in the lower >end cards as well since the entire code base is unified - >just not as dramatic as the higher end cards. I'm guessing >you have to be a developer to get this idea. There is /has been no problem with GF4 AF at 8/16!!! As the "great" game and benchmark scores bear this out. The AF problem is when 32/64 is enabled the Fillrate of a GF4 will drop through the roof way, way below a GF3, Your statement of "code opt that is good is good for all" is completely ill-relevant here, No one is looking for an AF improvement across the board for all the boards from the AF fix and none is expected, why? Because "all code opt is good for all" is not true in this case, Go back and read your benchmarks for OGL in the 23.11 non AF "improved" drivers on your GF2/3 and then fire up your AF "improved" 29.42s and run again OGL and you will not see any real improvement period.If it wasn

[Guest]

"The AF problem is when 32/64 is enabled the Fillrate of a GF4 will drop through the roof way, way below a GF3"Around and around we go - where will it stop - nobody knows!! LOL! :-lolARGGG MATEY! Read through this thread again. The original - and always friendly - disagreement was whether the v29 series is good for GF3 and GF2 users as well as a GF4 users (it started with the "RE: Friendly reminder-GF4 only..." post). MY argument I've been trying to get across is one and one only, and summed up by your reply:""code opt that is good is good for all" is completely ill-relevant here"ITS THE MOST RELEVANT. We're not (or weren't) discussing what is good, bad or needed in the GF4 only line. My second post to this thread pointed out that I jumped the gun and these weren't the DX fixed drivers aimed at the GF4. So all that was left to discuss was: are these drivers good for users of the GF2, GF3 and GF4, and should they be recommended as such... Which brings us back to: code optimizations in a unified driver architecture are good for all cards down the line - if not as much as the highest end where the real targeting takes place."No one is looking for an AF improvement across the board for all the boards from the AF fix and none is expected"Huh? EVERY driver is a potential candidate for use - either it performs better and fixes bugs or performs worse and has more bugs. Everyone is looking for an AF improvement (among any other improvements possible) across the board! And the v29 series provides some of those improvements. "Because "all code opt is good for all" is not true in this case"It absolutely IS true in this case on my machines and in my testing. OpenGL Aniso performance IS much, much improved on the GF4 - and even better - is also slightly improved on the GF3 and GF2. Now, if you are speaking Flight Sim only performance I completely agree (and maybe thats where we're getting hung up here - you are and I'm not). These drivers have no improvements in Flight Sim - but are no worse either. Its the OTHER games and apps that almost every FS user runs that get the improvements with the v29's bug-wise AND performance-wise. On all of the cards: GF2, GF3 AND GF4.Ah, my Cadillac awaits! Sweet wind in my hair baby... Sweet wind! Play on Mr. Page!:-)http://members.rogers.com/eelvish/elrondlogo.gifhttp://members.rogers.com/eelvish/flyurl.gif

[Guest PaulL01]

>ARGGG MATEY! Read through this thread again. The original >- and always friendly - disagreement was whether the v29 >series is good for GF3 and GF2 users as well as a GF4 users >(it started with the "RE: Friendly reminder-GF4 only..." >post). MY argument I've been trying to get across is one >and one only, and summed up by your reply: Elrond,I will leave it at this:I disagree with you, I said "no one is looking for AF improvement across the board" as to GF2/3 cards and that includes the NV driver delopment teem. IN general it is true you can hope for an improvement, but History has born out that once driver devlopment is focused on a newer card there >generaly:)Traded in the Caddy for older Lamborghini Muria.....zzZZZzzzZZZooooOOOOOO (2nd gear)zzZZZzzzoooOOOOOOooooOOOOOOOPaul