View Full Version : So now we know
bkswaney
05-24-03, 01:00 AM
I think these two things say a lot.
____________________________________________
Since NVIDIA is not part in the FutureMark beta program (a program which costs of hundreds of thousands of dollars to participate in) we do not get a chance to work with Futuremark on writing the shaders like we would with a real applications developer.
__________________________________________________ __
ATI stated:
The 1.9% performance gain comes from optimization of the two DX9 shaders (water and sky) in Game Test 4 . We render the scene exactly as intended by Futuremark, in full-precision floating point. Our shaders are mathematically and functionally identical to Futuremark's and there are no visual artifacts; we simply shuffle instructions to take advantage of our architecture. These are exactly the sort of optimizations that work in games to improve frame rates without reducing image quality and as such, are a realistic approach to a benchmark intended to measure in-game performance. However, we recognize that these can be used by some people to call into question the legitimacy of benchmark results, and so we are removing them from our driver as soon as is physically possible. We expect them to be gone by the next release of CATALYST.
__________________________________________________
As I see it Nvidia got ****ed by not staying with FM's beta program.
ATI worked hand in hand with FM to make sure there shader was at top performance for "there bench".
Nvidia did not. But "Game Companies" do work very close with nvidia. Even more so than ati.
That is why you see the FX's suxing on PS 2.0 3DM03 and kicking butt in games.
It's easy to read between the lines on this one just from what both companies have said.
edit by StealthHawk: dont circumvent the swear filter. Thanks.
nvidia also said...
"We don't know what they did, but it looks like they have intentionally tried to create a scenario that makes our products look bad."
i think not staying in the futuremark program was very bad for nvidia... that money is pocket change for them
reever2
05-24-03, 01:05 AM
Originally posted by bkswaney
That is why you see the FX's suxing on PS 2.0 3DM03 and kicking butt in games.
Until you start seeing PS 2.0 games, then Nvidia has to fix its **** architecture
bkswaney
05-24-03, 01:06 AM
Originally posted by jAkUp
nvidia also said...
"We don't know what they did, but it looks like they have intentionally tried to create a scenario that makes our products look bad."
i think not staying in the futuremark program was very bad for nvidia... that money is pocket change for them
I agree big time on that one.
If nvidia had stayed with FM they would not be here today.
They got left out in the cold.
They need to sux it up and go on.
Quite trying to cheat on 3DM03 and keep focas on the games.
They can pick back up with FM on future benchmarks.
bkswaney
05-24-03, 01:10 AM
Originally posted by reever2
Until you start seeing PS 2.0 games, then Nvidia has to fix its **** architecture
That might not be true.
It seems to be Nvidia's idea of 2.0. Most every game company is going to work with nvidia to give them what they need.
Heck with 60% of the market I see why.
Nvidia just did not match there PS up with futuremarks it seems.
Originally posted by jAkUp
nvidia also said...
i think not staying in the futuremark program was very bad for nvidia... that money is pocket change for them
have you ever heard what Nvidia says about 3dmark?
i know there are journalists outhere,that has backed 3dmark
as a reliable benchmark.. so maybe they are right..
but what if they are wrong and Nvidia is right?
would you still pay thousands of $$$ for SOmething you dont believe?
Since NVIDIA is not part in the FutureMark beta program (a program which costs of hundreds of thousands of dollars to participate in) we do not get a chance to work with Futuremark on writing the shaders like we would with a real applications developer
Interpreted: We did not get to have things done our way. Things were done according to dx9 spec and we didn't like that. You see, our cards don't perform so well when they run standard dx9 spec. We need for developers to write special code to get our cards to run acceptably. We didn't get our way so we left the program. Once we left the program we were forced to cheat. It's not our fault. We're a product of our environment and bad engineering decisions.
ATI worked hand in hand with FM to make sure there shader was at top performance for "there bench".
Interpreted more closely to the truth: FutureMark made sure they created the bench according to DX9 Standards. Ati, as a result of engineering their board according to dx9 specs, performs very well in a dx9 futuremark benchmark.
Nvidia did not. But "Game Companies" do work very close with nvidia. Even more so than ati.
That is why you see the FX's suxing on PS 2.0 3DM03 and kicking butt in games.
How many dx9 games have you seen? 3dmark is not the only dx9 benchmark out there. There are 2 more. Shadermark and Rightmark. Both of those also show the r300 destroying the nv30 using dx9 shaders. Games companies will HAVE to work with Nvidia or they will get poor performance in dx9 games.
Listen guys, I hear a lot of people bitching about how 3dmark is not fair to Nvidia. The only thing not fair to Nvidia is Nvidia themselves. They chose to engineer a board that does not perform well using standard dx9 coding. They chose to engineer a board that developers will have to optimize specifically for in order to get it to perform acceptably. Nvidia made some pretty bad decisions and now they have to deal with it. Ati on the other hand engineered a board that runs DX9 shaders great right from the gate. No special optimizations needed. This is not Futuremarks fault or ATI's fault that Nvidia felt they had to deviate from dx9 standards. Not all games companies are going to be willing to spend the extra time to optimize specifically for the nv30 to get it to run up to speed. I'm sure a lot will though. They won't have much choice if they want to sell their games to a wider audience.
You are right though. If Nvidia would have stayed in the Futuremark Beta they would not be in this position today. Not because the benchmark would have been programed any differently, but because they would have known about the developers version that allows you to go off the rails. They would have known they would be caught and not used that cheat, which turned out to be so easy to detect with the developers version.
Originally posted by Nv40
would you pay THousands of dollars for something that
YOU DONT BELIEVE?
have you ever heard what Nvidia says about 3dmark?
i know there are journalists outhere,that has backed 3dmark
as a reliable benchmark.. so maybe they are right..
but what if they are wrong and Nvidia is right?
would you still pay thousands of $$$ for SOmething you dont believe?
Originally posted by Nv40
would you pay THousands of dollars for something that
YOU DONT BELIEVE?
lol nvidia does... they just say they dont. why do you think they spent the time to optimize for it?? not to mention winning in 3dmark03 is a very big deal because alot of consumers go by that. nvidia wants to sell.
trust me.. thousands of dollars is nothing to nvidia. they would of done it if they felt their hardware accomodated it nicely... and at that time the nv30 and 3dmark03 did not agree
Heck with 60% of the market I see why.
Nvidia just did not match there PS up with futuremarks it seems.
Again. Nvidia did not match there PS up with the DX standard. This has nothing to do with Futuremark. Quit blaming futuremark for Nvidia's poor decisions.
How long do you think that 60% is going to last with these types of shenanigans? Market share doesn't change overnight. More and more of that market share is going to slide over to ATI. Do you really think developers will like having to write special code to get nv30 cards to perform acceptably with dx9 shaders? No doubt some of them will do it. They won't like spending the extra time that could be used to do other things, but they will do it. Not all will though. The more r300 cards ATI sell, the less developers will be willing to go out of there way to optimize for Nvidia. And ATI has been selling a heck of a lot of r300's. With the type of attitude Nvidia has showed lately, I'm sure more and more people will have ATI cards in there machines.
I'm using a 9700 now, but if ATI was using the tactics Nvidia has been using you can rest assured I wouldn't have it in my machine.
Originally posted by jAkUp
lol nvidia does... they just say they dont. why do you think they spent the time to optimize for it?? not to mention winning in 3dmark03 is a very big deal because alot of consumers go by that.
trust me.. thousands of dollars is nothing to nvidia. they would of done it if they felt their hardware accomodated it nicely... and at that time the nv30 and 3dmark03 did not agree
you have not answered my question JAck...
Nvidia can be millionaire ,but their money ,will not be invested in
things they dont believe..
so i ask YOU ,would you pay for something say $1 dollar or 1 million ,even if you have the money ,if you strongly feel
you dont have to do it ,for X,Y,Z reasons ,because it benefits
Nothing you ,and because you dont believe in it.?
i would. but i would also believe in it. because like i said, winning in 3dmark03 is not just a nice number to show... it sells more. you will probably double what you put into the futuremark program by just having a *legit* badass 3dmark03 score...
i dont agree with using 3dmark03 as a tool to choose what vcard you need... but millions of people do. since nvidia felt their card wasnt up to speed in 3dmark03, they dropped the program, then said "3dmark03 is useless" then we see them cheating to get good 3dmark03 scores... how strange....:rolleyes:
would you pay for something say $1 dollar or 1 millions
if you feel you dont have to , because you dont believe in it.?
Not if I knew I had millions of gullible people out there that believed every line I fed them. It's not about the money Nv40. It's about them looking bad because their card runs dx9 shaders poorly. They were in the beta program for most of the development process. They didn't have a problem with it until they got their nv30's and saw how poorly it ran. They chose the only option they thought they had. Denounce the benchmark that shows this. They can't exactly denounce it if they're still a beta member can they?
digitalwanderer
05-24-03, 01:48 AM
Originally posted by jjjayb
It's not our fault. We're a product of our environment and bad engineering decisions.
You raised a lot of excellent points most well, but the above is still making me giggle! :lol: :lol:
Originally posted by jjjayb
Not if I knew I had millions of gullible people out there that believed every line I fed them. It's not about the money Nv40. It's about them looking bad because their card runs dx9 shaders poorly. They were in the beta program for most of the development process. They didn't have a problem with it until they got their nv30's and saw how poorly it ran. They chose the only option they thought they had. Denounce the benchmark that shows this. They can't exactly denounce it if they're still a beta member can they?
and who have say it about money?
it because NVIDIA believes they DONT HAVE TO PAY.
and support something they dont agreed,or believe.
Microsoft is Billionare ,but that doesnt means they have to support
every one who ask $$$ to test their products ,in a test they
Dont believe. for X,Y reasons ..
dont you think is possible ,that Nvidia knows something about
3dmark2003 that you dont ? yes thats impossible.. i know.
but lets say you are the president at Nvidia
and lets say Nvidia is right ..about 3dmark in their claims
of ineficiency and biassed for other cards of the benchmark..
(i know we already have the guarantee of journalists about 3dmark reliability and integrity) beta-members of Futuremark.. BTW.
but lets suppose this is true. and that by accident or coincidence
Nvidia is right about 3dmark+ATI conpirancy.
(again, i know multibillionare companies in Busines$$ ,will
never take an oportunnity ,in an unfair way ,to decrease the sales of another one .. ) just ask microsoft.
so the question is ,Would you still pay Hundreds of thousands dollars for something you dont believe?
the answer is -> NO
but the opposite ,you will warn the public about the benchmark.
and will discredit it ,as a reliable test for their cards.
ChrisRay
05-24-03, 02:02 AM
I think thing is this.
ATi and Nvidia have very different visions about the future of shaders and how they will come to pass.
Currently it seems Microsoft seems to favor ATI's Vision, As ATI's vision is a part of Microsoft's standard.
Nvidia favors its own ideals on how shader aplications will be ran.
We need to diagnose the real problem here, Not who is right and who is wrong.
Nvidia cards perform very well under there own Compiled Enviroments, ATI cards run very well under Microsofts Standards.
Nvidias Hardware has been optimised very much for its own "vision" of how API's will handle shader instructions.
IN this Case, Nvidia calls upon partial, fragmented, and interger precision at specific times for rendering a given scene.
Under many circumstances this is probably acceptable. And this why the Nv30 hardware was designed as it was. Nvidia seems to be trying to hold onto it's old architecture as it moves towards future architecture,(you have to take note that Nvidia was very close with Microsoft when the DX 8.0 standard was created)
This would ensure ultimate performance in old/modern/ But not exactly future products. Assuming All future products conform to the DirectX standard.
However, If they do, And they go the Nv30 path, And choose to use its partial, fragmented, and integer precision. Then it would perform fine in future aplications. It's obvious that Nvidias Pixel Shader is meant to handle a great deal of various precisions which are not listed in DirectX 9.0 specifications
It would also seem Nvidia doesn't seem to want to give up Multitexturing quite just yet. Which is a good indication by their architectures, Which seem quite strong in this field. (Multi Texturing for todays and yesterdays game is quite relevent performance wise)
However I'm not quite sure how relevent multi texturing will be with shaders taking over the way textures are mapped.
Now We have ATI's stance, which is purely conforming to The DirectX 9.0 specification given to them by Microsoft. This specification is the very heart of ATI's design and it has developed a completely new architecture around it.
You can gather this by some of the way it handles its features, Such as Multi Sampling Anti Aliasing ect, Purely conforment with DX 9.0 specifications.
I actually see nothing wrong with this. However it has left some things to be desired with older forms of aplications. Which are more dependent on the T&L engine. And older DirectX 7.0/6.0 Aplications. And Less on Multi texturing. Where ATI cards have not quite been as powerful as the Nvidia counterparts.
However for the most part they are functioning correctly. But not always optimally.
In this case I would have to say ATI is looking "forward" with its technology in this respect, And trying harder not to hold onto its Legacy support, (I use legacy as a term loosely)
So here we have it. ATI is definately following Microsofts DirectX to the letter, (I am certain microsoft is loving this. As they really do like to control the standards for all things regarding PCs)
And then we have NVidia, Which is trying to develop its own standards for shader aplications.
Who's right in this issue I cannot say,Unfortunately it's going to be a wait and see scenerio, Will Nvidias PR relationship with Developers be there saving grace?
Will nvidias standard Be the standard for Which Shader games are played? Or will the standard Microsoft And ATI used be the future in tommorrows aplications? It should be noted that Nvidia does "conform" with the standard when it is forced too. Not exactly Yielding Nvidia product performance in the best of Light.
Now I know some of you must be thinking? Where does this Leave OpenGL? Which really doesn't have a "standard" thats forced upon everyone, Being completely open Source. Nvidia is pretty much free to do whatever they want with it.
In This case you'll see a great deal of OpenGL titles (obviously anything based off the doom 3 engine) Optimised for the Nvidia hardware and shader pathways.
Could we see a similar scenerio back in the days of 3dfx Where Nvidia really did destroy in OpenGL aplications due to its hardware being used in fullest? And only mediocre results in Direct3d compared to its competitors (3dfx)
I guess only time will tell at this point. So Who knows. Btw I apologise for any typos I might have made in advance and it should be noted everything I have written here is purely speculative.
bkswaney
05-24-03, 02:19 AM
Originally posted by ChrisRay
I think thing is this.
ATi and Nvidia have very different visions about the future of shaders and how they will come to pass.
Currently it seems Microsoft seems to favor ATI's Vision, As ATI's vision is a part of Microsoft's standard.
Nvidia favors its own ideals on how shader aplications will be ran.
We need to diagnose the real problem here, Not who is right and who is wrong.
Nvidia cards perform very well under there own Compiled Enviroments, ATI cards run very well under Microsofts Standards.
Nvidias Hardware has been optimised very much for its own "vision" of how API's will handle shader instructions.
IN this Case, Nvidia calls upon partial, fragmented, and interger precision at specific times for rendering a given scene.
Under many circumstances this is probably acceptable. And this why the Nv30 hardware was designed as it was. Nvidia seems to be trying to hold onto it's old architecture as it moves towards future architecture,(you have to take note that Nvidia was very close with Microsoft when the DX 8.0 standard was created)
This would ensure ultimate performance in old/modern/ But not exactly future products. Assuming All future products conform to the DirectX standard.
However, If they do, And they go the Nv30 path, And choose to use its partial, fragmented, and integer precision. Then it would perform fine in future aplications. It's obvious that Nvidias Pixel Shader is meant to handle a great deal of various precisions which are not listed in DirectX 9.0 specifications
It would also seem Nvidia doesn't seem to want to give up Multitexturing quite just yet. Which is a good indication by their architectures, Which seem quite strong in this field. (Multi Texturing for todays and yesterdays game is quite relevent performance wise)
However I'm not quite sure how relevent multi texturing will be with shaders taking over the way textures are mapped.
Now We have ATI's stance, which is purely conforming to The DirectX 9.0 specification given to them by Microsoft. This specification is the very heart of ATI's design and it has developed a completely new architecture around it.
You can gather this by some of the way it handles its features, Such as Multi Sampling Anti Aliasing ect, Purely conforment with DX 9.0 specifications.
I actually see nothing wrong with this. However it has left some things to be desired with older forms of aplications. Which are more dependent on the T&L engine. And older DirectX 7.0/6.0 Aplications. And Less on Multi texturing. Where ATI cards have not quite been as powerful as the Nvidia counterparts.
However for the most part they are functioning correctly. But not always optimally.
In this case I would have to say ATI is looking "forward" with its technology in this respect, And trying harder not to hold onto its Legacy support, (I use legacy as a term loosely)
So here we have it. ATI is definately following Microsofts DirectX to the letter, (I am certain microsoft is loving this. As they really do like to control the standards for all things regarding PCs)
And then we have NVidia, Which is trying to develop its own standards for shader aplications.
Who's right in this issue I cannot say,Unfortunately it's going to be a wait and see scenerio, Will Nvidias PR relationship with Developers be there saving grace?
Will nvidias standard Be the standard for Which Shader games are played? Or will the standard Microsoft And ATI used be the future in tommorrows aplications? It should be noted that Nvidia does "conform" with the standard when it is forced too. Not exactly Yielding Nvidia product performance in the best of Light.
Now I know some of you must be thinking? Where does this Leave OpenGL? Which really doesn't have a "standard" thats forced upon everyone, Being completely open Source. Nvidia is pretty much free to do whatever they want with it.
In This case you'll see a great deal of OpenGL titles (obviously anything based off the doom 3 engine) Optimised for the Nvidia hardware and shader pathways.
Could we see a similar scenerio back in the days of 3dfx Where Nvidia really did destroy in OpenGL aplications due to its hardware being used in fullest? And only mediocre results in Direct3d compared to its competitors (3dfx)
I guess only time will tell at this point. So Who knows. Btw I apologise for any typos I might have made in advance and it should be noted everything I have written here is purely speculative.
You saved me a lot
of work. :D
That's about what I've been thinking. Wrong or right. :ass:
This is a big reason I'm going to hold off on buying any new card for a bit.
Yes the NV30 and up will rock in OGL. :nana:
DOWN WITH MS OPEN SOURCE IS THE WAY!!!!!!!!!!!!!! :super: :angel3:
Originally posted by bkswaney
You saved me a lot
of work. :D
That's about what I've been thinking. Wrong or right. :ass:
:super: :angel3:
yes ChrisRay have many really Good points ..
but also i will like add ,that in games TWO video cards ,from diferent
companies that are 100% directx9 compatible , to the letter.. as you say it. can have a huge diferences in performance in games..
even if both follow Microsoft . why?
because there can be also other hardware diferences ,like
PIpelines organizations,Fillrate ,bandwidt and other hardware diferences ,
that can make an aplication or a game benefit more the design of
X company than the design of Y company..
in other words the same GAME or benchmark (3dmark)
can be written in many ways ,with exactly the same IQ and even better,
but with the performance switching in favor card#1 over Card#2 and in the other way.
it all depends of the game developer choices, efficiency.
and at the worst their integrity. good game developers will program
in a balanced way . the best IQ/performance for both cards.
Design 3dmark2003 with heavy use of multitexturing ,heavy use of NVidia PS1.x ,or 2.0+ (MS standar) and force ATI to multipass at all times,what Nvidia can do in just ONE PASS ,and you will see ATI claiming the benchmark is not reliable.. :)
Very well thought out and well spoken post Chris. You pretty much explained the situation perfectly.
I think Nvidia should have focused less on the older and modern products and more on the future products. They are selling the cards for $500 afterall. It's not like you're just going to throw the card away after 6 months. Even if Nvidia want's to call upon partial, fragmented, and interger precision at specific times for rendering a given scene, they should make sure the full precision is up to snuff first. Basically, they shouldn't use the partial precision to be "as good as, or slightly better than" anyone else using full precision. They should have targeted the full precision to be "as good as" or better than their competitors and the partial to completely blow the competition away.
I think it's great Nvidia went above and below the standard. But it's bad that they don't have the performance when going above the standard or even meeting the standard. The above the standard doesn't do much good if it runs too slow to implement.
I don't think so much that they made a mistake with this architecture. It's more because the r300 that it seems like a failure. Because they were not expecting ATI to implement as well as they did when they did they didn't feel they needed to push the bar so high . Who really expected ATI to pull of what they did with the r300? I certainly didn't. I'm sure they weren't expecting it either. Not with ATI's past record on releasing cards. Especially when they heard ati was using .15 micron. Most of the "experts" didn't think they'd be clocked much over 225mhz on 15 micron with all the dx9 features packed in there.
If ATI hadn't released the r300 we would all be in agreement that the nv30 is a kick ass card. But after the 9700 the nv30 was a dissappointment. Especially for anyone used to Nvidia being on the top. I was expecting alot more from them after seeing the 9700. But I guess that wasn't realistic. If the r300 was never released, it would be the pahrelia against the nv30. Then the nv30 would have looked like a whopper.
I think if the r300 was never released, or only performed on the same level as the pahrelia, nvidia never would have had a problem with 3dmark03. They would have used it's marketing value for all it's worth. Just like they did with 3dmark2001. It's not that 3dmark03 makes nvidia looks bad. It's that the r300 makes the nv30 look bad.
What bothers me is the way Nvidia has reacted to all of this though. Rather than sucking it up and moving on, they have gone to slimeball tactic mode.
Look at the nv30. They used reduced quality in the drivers on release. Some of this quality has only now been cleared up. This is Nvidia, the company with great drivers. I don't feel for one bit that the reduced quality was an accident. It served it's purpose. It got them better fps in games when the card was initially reviewed. The misleading control panel sliders. Suggesting review sites use control panel settings that they now darn well are not comparable when benching against the competition. Flaming futuremark rather than just dealing with not being first for once. Then after flaming futuremark, cheating in the benchmark they said they didn't like in the first place. Then after being caught, blaming futuremark again. It all just sits really bad with me.
These tactics are going to lose Nvida more customers than just having a slower card would have. After they saw the r300 they should have just accepted that they weren't going to have the fastest card for a few months. They should have just dealt with being slower in 3dmark. They should have kept overall image quality perfect, even if it meant running slower (in games too) and said hey, we'll work harder on the next card. If they'd done that, I would have actually considered buying their next card. As it is now, know way. They have a lot of confidence to regain before I'll do that. Ati is no saint either, but i'll take the lesser of two evils. I do feel ATI is actually getting better in the public image department though while Nvidia is getting worse.
Nvidia really needs to clean up their image. Die hard Nvidia fans will ignore the tactics of late, but the average joe won't. Alot of former Nvidia fans have already jumped ship to get the r300. These kinds of things will only make them less likely to come back. And I've seen enough people on the web today who have Nvidia cards now say they won't buy another one after this fiasco. Sure, it's only a benchmark. But it's also your image.
Just my 1 a.m. rambling thoughts.
I think Nvidia should have focused less on the older and modern products and more on the future products.
would you explain that more?
if ATI is more focused with the future?
the why John carmack use the Nv3x for their next game ,
next to Doom3 and maybe final one..?
have you ever read ,when JC told.. "i have already reached the limits of R300" not in the Nv30. did you know that NVdia hardware PS/VS is a hair
close to microsoft PS/VS 9.1?
On a different subject ( not related ;) ) something should be done about the APIs. The problems with the two current ones are
1) DirectX - MS Control and not cross platform . But has a fairly fixed standard.
2) OpenGL - Not under some company's control but has the propreitary extensions issue which makes it hell for game devs to optimize for a certain platform.
Basically we need a new standard ( maybe opengl 2.0 can address this ) which incorporates good points from both and thus game devs have a fairly fixed, cross platform, free standard to target to.
Then game devs can say to hell with MS :angel:
My take is this , Nvidia tried to deliver a knock out blow to ATI with CG and proprietary extensions. They wanted to be in the same relative position Microsoft or Intel is to their competition. Basically the tried to leverage there marketshare a little too early in my opinion and tried to push the industry in a direction that was favorable to them. Alas the R300 proved to be a brick wall and Nvidia broke it self apart on it.
I think some ppl. don't realise how much trouble Nvidia is really in. They are loaded with debt and liabilities. There product line is a flop. As far as optimising DX9 for the FX because of marketshare , keep dreaming. Nvidias marketshare lead can be attributed to the GF4mx which is DX7 . I don't think they will be coding for that. They will code for the most prevalent DX9 compatible hardware which at this point is the r300 series.
have you ever read ,when JC told.. "i have already reached the limits of R300" not in the Nv30. did you know that NVdia hardware PS/VS is a hair
close to microsoft PS/VS 9.1?
The question is, Have you read it? Actually, I'm sure you have. But, Do you understand it?
Quote from john carmack:
For developers doing forward looking work, there is a different tradeoff --
the NV30 runs fragment programs much slower, but it has a huge maximum
instruction count. I have bumped into program limits on the R300 already.
This is not with code that is actually going into the game. With the instruction lengths he was using it would run like a slideshow on the nv30.
By the way, have you looked at the 9800? It actually allows for MORE instructions than the nv30 or the nv35. It is a hair and a half closer to ps/vs 9.1 than the nv30 and the nv35 ;-)
bkswaney
05-24-03, 03:44 AM
I think a lot of it boils down to Nvidia using 16 and 32 precision only and not adding 24 like ati did.
It's killing Nvidia trying to use Full 32 precision against ati's 24.
They just cannot get the performance mark in 32 up to ati's 24.
One things for sure. It's no wonder Mr Nvidia did not get any bonus. NV has made some bad choices over the past year.
Starting by not putting a 256bit memory controler on the NV30.
They need to get there head out of the rain for sure.
StealthHawk
05-24-03, 03:57 AM
Originally posted by ChrisRay
IN this Case, Nvidia calls upon partial, fragmented, and interger precision at specific times for rendering a given scene.
A very nice post. I do take issue with the above quoted statement though.
NV35 dropped Integer support from the shader pipeline.
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.