Go Back   nV News Forums > Hardware Forums > Benchmarking And Overclocking

Newegg Daily Deals

Reply
 
Thread Tools
Old 05-26-03, 09:17 PM   #109
StealthHawk
Guest
 
Posts: n/a
Default

Quote:
Originally posted by Behemoth
lets investigate further on your example:

1) 2+2=4
2) 2^2=4


assume operation ^ is PS2 shader
assume operation + is IHV siturational shader
assume both integers on the left side are inputs.
assume the integer on right side is output.

futuremark is testing PS2 shader performance of a video card, since the driver maker knows in this particular case, with input 2,2, both shaders are mathematically functional outputly identical, if the driver maker substitute a IHV siturational shader +, do you think it is a cheat?
Yes, I think that is a cheat. And I referred you to CatalystMaker's comments. He said that is NOT what they are doing. They are NOT optimizing for specific cases of a function, but optimizing the WHOLE function.

If you think differently, you are saying he is a liar, which is fair enough. You are entitled to your opinion. But either ATI is lying, and they are cheating because they are not doing what they said, or they are not lying and they are not cheating because they are doing what they said. Those are the only 2 possibilities.
  Reply With Quote
Old 05-27-03, 12:03 AM   #110
Behemoth
radeon 9800 pro
 
Behemoth's Avatar
 
Join Date: Apr 2003
Location: Darkness Falls
Posts: 841
Default

Quote:
Originally posted by StealthHawk
Yes, I think that is a cheat. And I referred you to CatalystMaker's comments. He said that is NOT what they are doing. They are NOT optimizing for specific cases of a function, but optimizing the WHOLE function.

If you think differently, you are saying he is a liar, which is fair enough. You are entitled to your opinion. But either ATI is lying, and they are cheating because they are not doing what they said, or they are not lying and they are not cheating because they are doing what they said. Those are the only 2 possibilities.
i would wait to see what futuremark would say after their investigation
Behemoth is offline   Reply With Quote
Old 05-27-03, 12:34 AM   #111
Behemoth
radeon 9800 pro
 
Behemoth's Avatar
 
Join Date: Apr 2003
Location: Darkness Falls
Posts: 841
Default

Quote:
Man Rage3D sure is a popular site these days! Thought I would pop in for a couple of points CM style!

1) I have nothing to say about what our competition has done. You guys can decide and hopefully the truth which is in front of you wont be overlooked

2) Looks like Beyond3D has already done my work for me. I quote two items. First from our guy Chris Evenden " 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."

That is a very important statement you need to understand. When every CATALYST comes out and you see performance boosts, what do you think you are getting? To put it bluntly it is optimizations which make the original code run more efficient.That is how we can continually improve performance. Usually it is applicable to all situations but this one in question was in particular to 3DM (as no other games use its engine).

The second comment I would like to relate that to is from Tim Sweeney, the lead developer for the UT games. (also found on Beyond3d)

".. any code optimization performed on a function that does not change the resulting value of the function for any argument, is uncontroversially considered a valid optimization. "

That guys is exactly what we did. We optimized the code without changing the end resutl.

3) I had a few meetings today and it was decided that we back out the 3DMark optimizations. I was not in favour of this as we are not cheating nor degrading image quality. We are giving ATI users the benefit of our research on the code. Anyways it was decided that we back out the optimizations but I promise you we will continue to work to find ways of getting the percentages back!

So there you have it. A little explanation goes a long way to understand this complex industry.

As anyone who knows me from the threads will tell you, I am a straight up guy, and if ATI was up to funny business I would not be associated with them any longer.

Now get back to some gaming on your RADEON's or "no drivers for you"

PS. Althought our guy Chris promised we would back out the optimizations from 3DMark in our next CATALYST, I will only permit that if it doesnt delay our postings. You have my word we will continue to post at the feverish rate we started last year, that left our competition in our dust!

Later everyone

CM
Quote:
Usually it is applicable to all situations but this one in question was in particular to 3DM
"Usually" means not always.
"this one in question was in particular to 3DM" means he is refering to 3DM only when he talks about all situations, he already hinted its situarational, was particular in 3DM, i can not understand his statements any better in any other way.
then
Quote:
".. any code optimization performed on a function that does not change the resulting value of the function for any argument, is uncontroversially considered a valid optimization. "

That guys is exactly what we did. We optimized the code without changing the end resutl.
right, any argument can mean any argument, but only in particular to 3DM , because he said this one in question was in particular to 3DM

Quote:
Our shaders are mathematically and functionally identical to Futuremark's and there are no visual artifacts
hey if they are the same, who expect there is any visual artifacts? futuremark suspected ati not using PS2 shaders not suspected ati's PS2 shaders have bugs! big difference.

Quote:
We optimized the code without changing the end resutl.
why didnt he just say "we optimized the shaders"?
i will tell you, because if they got caught later it doesnt work in all siturations, they could explain "we didnt say it works in all siturations, it just doesnt change the end result in particular to 3DM"

this guy is unbelievable, he certainly was not lying, but i understand him perfectly.
Behemoth is offline   Reply With Quote
Old 05-27-03, 01:04 AM   #112
Sazar
Sayonara !!!
 
Sazar's Avatar
 
Join Date: Jan 2003
Location: Austin, Texas
Posts: 9,297
Default

Quote:
Originally posted by Behemoth
"Usually" means not always.
"this one in question was in particular to 3DM" means he is refering to 3DM only when he talks about all situations, he already hinted its situarational, was particular in 3DM, i can not understand his statements any better in any other way.
then

right, any argument can mean any argument, but only in particular to 3DM , because he said this one in question was in particular to 3DM


hey if they are the same, who expect there is any visual artifacts? futuremark suspected ati not using PS2 shaders not suspected ati's PS2 shaders have bugs! big difference.


why didnt he just say "we optimized the shaders"?
i will tell you, because if they got caught later it doesnt work in all siturations, they could explain "we didnt say it works in all siturations, it just doesnt change the end result in particular to 3DM"

this guy is unbelievable, he certainly was not lying, but i understand him perfectly.
he is saying what he is saying because that is what is being done..

read up about it on b3d forums where the guys have basically broken everything down and explained it as best as is possible...

concerning the optimizations... it is the code you optimize thats where the performance comes from basically... ati has already specified WHAT it optimized so I see no reason for you to pick on this point and overanalyze it... it has been stated already...

concerning the bugs... he has put that AFAIk in reference to nvidia also appearing to render everything with full precision but not actually rendering 'everything' and certainly with anomalies that are not expected...

coding can be done in many ways to achieve the same result... if you have worked with coding as I take it you have... you will know what I am talking abouit...

you can most often times... especially with long strings of code find better ways to work something... it is expected simply because there are things one finds out... new paths or whatever it may be... to make the overall load less while @ the same time doing EXACTLY the same work... I use the word efficiency

nvidia's drivers are not doing the SAME work though... they are CHANGING the workload...

ati's stance though... about removing said application specific optimizations... I can fully understand and it does show character to step up.. pinpoint exactly what you have done and why and then state that you are removing it... thus far nvidia has yet to come out with anything close to resembling a statement that explains what is going on in their test results even though it is blatantly obvious...
Sazar is offline   Reply With Quote
Old 05-27-03, 02:01 AM   #113
Behemoth
radeon 9800 pro
 
Behemoth's Avatar
 
Join Date: Apr 2003
Location: Darkness Falls
Posts: 841
Default

Quote:
Originally posted by Sazar
he is saying what he is saying because that is what is being done..
i know

Quote:

read up about it on b3d forums where the guys have basically broken everything down and explained it as best as is possible...
fine, i probably dont need their interpretation.

Quote:

concerning the optimizations... it is the code you optimize thats where the performance comes from basically... ati has already specified WHAT it optimized so I see no reason for you to pick on this point and overanalyze it... it has been stated already...
i dont care if you found my interpretation pointless.

Quote:

concerning the bugs... he has put that AFAIk in reference to nvidia also appearing to render everything with full precision but not actually rendering 'everything' and certainly with anomalies that are not expected...
what bug?

Quote:

coding can be done in many ways to achieve the same result... if you have worked with coding as I take it you have... you will know what I am talking abouit...
yes i know how to program.

Quote:

you can most often times... especially with long strings of code find better ways to work something... it is expected simply because there are things one finds out... new paths or whatever it may be... to make the overall load less while @ the same time doing EXACTLY the same work... I use the word efficiency
right, i would call it better implementation.

Quote:

nvidia's drivers are not doing the SAME work though... they are CHANGING the workload...
ok, i also suspect ati's driver not doing the same work too, the sky and water shaders have been partially hard coded or pre-calculated.

Quote:

ati's stance though... about removing said application specific optimizations... I can fully understand and it does show character to step up.. pinpoint exactly what you have done and why and then state that you are removing it... thus far nvidia has yet to come out with anything close to resembling a statement that explains what is going on in their test results even though it is blatantly obvious...
when someone refuses to disprove his suspicous cheat but only chooses to not do it again, it is also blatantly obvious he was probably cheating.
Behemoth is offline   Reply With Quote
Old 05-27-03, 07:03 AM   #114
Grrrpoop
Wey aye man!
 
Grrrpoop's Avatar
 
Join Date: Jan 2003
Location: Newcastle, UK
Posts: 162
Default

Is nVidia's "implementation" better in terms of speed? Undoubtedly
Does it output the same quality that FM intended? No way.

Quote:
Originally posted by Behemoth
ok, i also suspect ati's driver not doing the same work too, the sky and water shaders have been partially hard coded or pre-calculated.
Your suspicions are completely unfounded as it has already been shown that ATI is resequencing the shader so it can be processed faster, whilst not actually changing it. The quality is exactly as FM intended. That is not pre-calculation. nVidia pre-calculated, re-wrote and degraded the shader quality. That's the difference.

Also, the same optimisations could be used in games, admittedly that would require each game to be recognized by the drivers, but if the drivers intercepted the shader calls then I'm sure these same optimisations could enhance the gaming experience whilst nVidia would have to rewrite shaders and reduce Iq to apply their "optimisation" to real games.

Quote:
Originally posted by Behemoth
when someone refuses to disprove his suspicous cheat but only chooses to not do it again, it is also blatantly obvious he was probably cheating.
Or they don't want the negative PR nVidia is now facing.
Besides, ATI outlined exactly what they did, and said that if ppl didn't like it then fine, it's not in the next driver. That's not the same as "refusing to disprove" a cheat. They acknowledge that something they did could be percieved as a wrongdoing and rather than labour the point have backed down.

By contrast nVidia have thrown accusations around and claim it's a conspiracy to discredit their products by FM.

Who has more credibility?
__________________
Don't be Care Less with your language
Grrrpoop is offline   Reply With Quote
Old 05-27-03, 07:29 AM   #115
zakelwe
Registered User
 
Join Date: Dec 2002
Posts: 768
Default

I will be happy for nvidia and Ati to optimise for 3dmark when and only when they have optimised for all other games

Otherwise they should leave it well alone.

Ati can say, well we do that for games, which is true, but they haven't done it for all games but instead select a benchmark program instead . Why ? Well, the same reasons nVidia did, but nVidia simply did it a lot worse.

Regards

Andy
zakelwe is offline   Reply With Quote
Old 05-27-03, 07:33 AM   #116
zakelwe
Registered User
 
Join Date: Dec 2002
Posts: 768
Default

Sorry, just reread that and didn't mean to indicate that nvidia just optimised worse, what I meant was they would have optimised like Ati but could not make up the difference, so resorted to cheats rather than optimisations.

Does that make sense

My view is we should not have either.

Regards

Andy
zakelwe is offline   Reply With Quote

Old 05-27-03, 09:57 AM   #117
Behemoth
radeon 9800 pro
 
Behemoth's Avatar
 
Join Date: Apr 2003
Location: Darkness Falls
Posts: 841
Default

Quote:
Originally posted by Grrrpoop
Is nVidia's "implementation" better in terms of speed? Undoubtedly
Does it output the same quality that FM intended? No way.
true.


Quote:
Originally posted by Grrrpoop
Your suspicions are completely unfounded as it has already been shown that ATI is resequencing the shader so it can be processed faster, whilst not actually changing it. The quality is exactly as FM intended. That is not pre-calculation. nVidia pre-calculated, re-wrote and degraded the shader quality. That's the difference.
having better ordering codes ati wouldnt want to use is totally unfounded.
the "shuffle instructions" codes dont work on places other than GT4 sky and water makes perfect sense to me, otherwise ati needs not purposely detect GT4 sky and water and alter/shuffle instructions in those particular cases, instead of simply use the better ordering PS2 codes in driver.
the equality of visual quality in GT4 doesnt neccessarily mean ati is using a general PS2 codes, which is what futuremark is trying to measure in GT4.


Quote:
Originally posted by Grrrpoop
Also, the same optimisations could be used in games, admittedly that would require each game to be recognized by the drivers, but if the drivers intercepted the shader calls then I'm sure these same optimisations could enhance the gaming experience whilst nVidia would have to rewrite shaders and reduce Iq to apply their "optimisation" to real games.
one can use valid optimization to cheat. i have cheated an automatic program marking system by hard coding all pre-calculated solutions to all possible known inputs into my program, result was full mark! and its a cheat! furturemark is testing valid PS2 performance, not only valid outputs.


Quote:
Originally posted by Grrrpoop
Or they don't want the negative PR nVidia is now facing.
Besides, ATI outlined exactly what they did, and said that if ppl didn't like it then fine, it's not in the next driver. That's not the same as "refusing to disprove" a cheat. They acknowledge that something they did could be percieved as a wrongdoing and rather than labour the point have backed down.
this "they dont want negative PR" is totally unfounded, if ati use the uber "shuffle instructions" codes in next driver, i dont think anyone would mind 8% more performance boost. The thing is the uber "shuffle instructions" codes would only work on GT4 sky and water, of coz imho.


Quote:
Originally posted by Grrrpoop
By contrast nVidia have thrown accusations around and claim it's a conspiracy to discredit their products by FM.

Who has more credibility?
this is tough question, i dont have clear answer yet.

Last edited by Behemoth; 05-27-03 at 10:39 AM.
Behemoth is offline   Reply With Quote
Old 05-27-03, 10:22 AM   #118
Behemoth
radeon 9800 pro
 
Behemoth's Avatar
 
Join Date: Apr 2003
Location: Darkness Falls
Posts: 841
Default

Quote:
Originally posted by zakelwe
I will be happy for nvidia and Ati to optimise for 3dmark when and only when they have optimised for all other games

Otherwise they should leave it well alone.

Ati can say, well we do that for games, which is true, but they haven't done it for all games but instead select a benchmark program instead . Why ? Well, the same reasons nVidia did, but nVidia simply did it a lot worse.

Regards

Andy
nvidia did it not that bad, i havent heard anyone spotted the difference before futuremark
the problem is futuremark is testing the general baseline PS2 performance, not testing how fast PS2 precision sky and water shaders can go with all sorts of scene specific optimizations. who knows if there are 99% of PS2 shaders ati cant even optimize by "shuffle instructions".
Behemoth is offline   Reply With Quote
Old 05-27-03, 01:24 PM   #119
Hanners
Elite Bastard
 
Hanners's Avatar
 
Join Date: Jan 2003
Posts: 984
Default

Quote:
Originally posted by Behemoth
nvidia did it not that bad, i havent heard anyone spotted the difference before futuremark
Not that bad? I would say cheating to the extent that it inflates your score by 25% is almost the definition of bad!
__________________
Owner / Editor-in-Chief - Elite Bastards
Hanners is offline   Reply With Quote
Old 05-27-03, 05:58 PM   #120
StealthHawk
Guest
 
Posts: n/a
Default

Quote:
Originally posted by Behemoth
having better ordering codes ati wouldnt want to use is totally unfounded.
the "shuffle instructions" codes dont work on places other than GT4 sky and water makes perfect sense to me, otherwise ati needs not purposely detect GT4 sky and water and alter/shuffle instructions in those particular cases, instead of simply use the better ordering PS2 codes in driver.
the equality of visual quality in GT4 doesnt neccessarily mean ati is using a general PS2 codes, which is what futuremark is trying to measure in GT4.
You keep saying this and it doesn't make an inkling of sense to me. We already know that ATI detects 3dmark03 and then optimized the shader routines in GT4. The sky shader and water shader optimizations only work in those specific situations. Of course they do, they are specific programs. How and why would they work anywhere else?

It's like saying that an optimization for Quake3 should work in UT2003, when they are different programs.
  Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
"GPU has fallen off the bus" error on 650M unless a CUDA program run first amonakov NVIDIA Linux 0 06-18-12 06:34 PM
Getting the proprietary nvidia driver to run with Debian 3.0 r0 (woody) Katchina404 NVIDIA Linux 9 01-12-03 08:49 AM
On The 333 Boards Does Anyone Run At 166fsb john19055 CPUs, Motherboards And Memory 22 07-30-02 07:39 PM

All times are GMT -5. The time now is 05:46 PM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.