Help me understand Kyle

[Quitch]

I have been re-hashing the issue with Kyle and someone from AVAULT.COM in this thread, if anyone wants to join the fun ..........................
http://www.hardforum.com/showthread.php?s=&threadid=688550&perpage=15&pagenumber=1

I've never seen so much drivel in my whole life. I do love the comments like "I looks to me like the Unified Compiler is being disabled by the patch..." Looks to me like someone is quoting press releases and passing them off as opinion

 

[AndY1]

Boy, is gonna feel Kyle stupid tonight:

http://www.xbitlabs.com/news/video/display/20031114041519.html

Luciano Alibrandi, European Product PR Manager for NVIDIA Corporation, has made a correction in regards previous information about NVIDIA’s Unified Compiler and 3DMark03 benchmark after getting into details with the company’s engineers. Apparently, the statement claiming that NVIDIA’s Unified Complier deployed to optimize pixel shader performance is disabled by the new version of 3DMark03 is not fully correct.


“I would like to inform you that a part of my response was not accurate. I stated that the compiler gets disabled, by 3DMark and that is in fact not true,â€￾ he said.

So, after all NVIDIA denied the problems between the Unified Compiler technology and the latest version of popular 3DMark03 benchmark. As a result, we may now conclude that the accusations in Futuremark direction from Hans-Wolfram Tismer, a Managing Director for Gainward Europe GmbH were not correct at all.

In October 2003 Santa Clara, California-based NVIDIA Corporation introduced its Unified Compiler integrated in its ForceWare 52.16 drivers to optimize Pixel Shader code for NVIDIA GeForce FX architecture in an attempt to improve performance of graphics cards powered by NVIDIA’s latest GPUs in variety of demanding applications.

NVIDIA said that the Unified Compiler technology tunes DirectX 9.0 execution on the GeForce FX GPUs, and can be used to correct any similar conflict that arises with future APIs. NVIDIA indicated the Unified Compiler as an automatic tuning tool that optimizes Pixel Shader performance in all applications, not just on specific ones. Officials from NVIDIA again stressed today that one of the things the Unified Compiler does is to reinstruct the order of lines of code in a shader. By simply doing this the performance can increase dramatically since the GeForce FX technology is very sensitive to instruction order. So, if the re-ordering is not happening NVIDIA’s GeForce FX parts have a performance penalty.

Since the complier is still active with the new version of 3DMark03 there is currently no explanations for performance drops of certain GeForce FX parts in the latest build 340 of the famous 3DMark03.

“The only change in build 340 is the order of some instructions in the shaders or the registers they use. This means that new shaders are mathematically equivalent with previous shaders. A GPU compiler should process the old and the new shader code basically with the same performance,â€￾ said Tero Sarkkinen, Executive Vice President of Sales and Marketing for Futuremark Corporation – the developer of 3DMark03 application.

He was indirectly confirmed by an ATI official yesterday, who said: “ATI has had a compiler since CATALYST 3.6. We did not have any problems with Futuremark’s changes.â€￾

 

[Sxotty]

long even for walt

Walt Unreal Tournament is not trying to foil the driver into thinking it is toasting buns with emma, so changing the footprint doesn't matter, just think a little bit about it and you will see that it will not be a problem, unless a developer tries specifically to hide what application is running which they won't it would be stupid for them to make the game run poorly on purpose.

 

[digitalwanderer]

Boy, is gonna feel Kyle stupid tonight:

http://www.xbitlabs.com/news/video/display/20031114041519.html

Luciano Alibrandi, European Product PR Manager for NVIDIA Corporation, has made a correction in regards previous information about NVIDIA’s Unified Compiler and 3DMark03 benchmark after getting into details with the company’s engineers. Apparently, the statement claiming that NVIDIA’s Unified Complier deployed to optimize pixel shader performance is disabled by the new version of 3DMark03 is not fully correct.


“I would like to inform you that a part of my response was not accurate. I stated that the compiler gets disabled, by 3DMark and that is in fact not true,†he said.

So, after all NVIDIA denied the problems between the Unified Compiler technology and the latest version of popular 3DMark03 benchmark. As a result, we may now conclude that the accusations in Futuremark direction from Hans-Wolfram Tismer, a Managing Director for Gainward Europe GmbH were not correct at all.

In October 2003 Santa Clara, California-based NVIDIA Corporation introduced its Unified Compiler integrated in its ForceWare 52.16 drivers to optimize Pixel Shader code for NVIDIA GeForce FX architecture in an attempt to improve performance of graphics cards powered by NVIDIA’s latest GPUs in variety of demanding applications.

NVIDIA said that the Unified Compiler technology tunes DirectX 9.0 execution on the GeForce FX GPUs, and can be used to correct any similar conflict that arises with future APIs. NVIDIA indicated the Unified Compiler as an automatic tuning tool that optimizes Pixel Shader performance in all applications, not just on specific ones. Officials from NVIDIA again stressed today that one of the things the Unified Compiler does is to reinstruct the order of lines of code in a shader. By simply doing this the performance can increase dramatically since the GeForce FX technology is very sensitive to instruction order. So, if the re-ordering is not happening NVIDIA’s GeForce FX parts have a performance penalty.

Since the complier is still active with the new version of 3DMark03 there is currently no explanations for performance drops of certain GeForce FX parts in the latest build 340 of the famous 3DMark03.

“The only change in build 340 is the order of some instructions in the shaders or the registers they use. This means that new shaders are mathematically equivalent with previous shaders. A GPU compiler should process the old and the new shader code basically with the same performance,†said Tero Sarkkinen, Executive Vice President of Sales and Marketing for Futuremark Corporation – the developer of 3DMark03 application.

He was indirectly confirmed by an ATI official yesterday, who said: “ATI has had a compiler since CATALYST 3.6. We did not have any problems with Futuremark’s changes.â€

Thanks for the quote/post AndY1, it's been the high point of my morning so far!

EDITED BITS: Just a bit more....

It's "asinine", not "assinine", "assignine", "assanine", etc.

Just wanted to get that off my chest.

We now return to our scheduled programming...

Thanks, and my apologies for annoying with my atrocious spelling.

Do you throw away your hammer because you cant use it to put in screws?

Just a quick technical note-o-correction: You can use your hammer to put in screws, you just have to hit a lot harder.

 

[Dio]

Yes, a carpet man did that in my house once. Didn't realise until I had to take the little f***ers out.

 

[WaltC]

long even for walt

Walt Unreal Tournament is not trying to foil the driver into thinking it is toasting buns with emma, so changing the footprint doesn't matter, just think a little bit about it and you will see that it will not be a problem, unless a developer tries specifically to hide what application is running which they won't it would be stupid for them to make the game run poorly on purpose.

I don't think you understood the point... The point is that since nVidia's "unified compiler" appears not to exist, at least as described by nVidia PR, and that what is actually occurring in the nVidia drivers is the same old application-specific optimizations nVidia's defended all year long (which depend on the ability of the drivers to detect the application being run in order to implement a specific environment strictly for that application), that whenever a game dev releases a patch which changes memory footprint or its internal shader code, nVidia's drivers will suffer exactly as they did when FM issued its patch.

Generally speaking, when a game developer makes such changes in a patch, he does so to make the game run better than it did before. But the *reason why* such a patch is made is completely irrelevant. FM obviously made the change to support one of its guidelines, to which nVidia and ATi and any other interested IHV have agreed, that driver detection of 3dmk03 is prohibited. So while FM made the changes to support its guidelines and maintain control of its software, a game developer might well make similar changes in order to improve his software. But in either case, application-specific optimizations which depend on driver detection of the application will cease to function.

It has nothing whatever to do with the idea that a dev is "deliberately trying to make his game run slower and less efficiently," which, when you think about it, is a highly unlikely approach for a dev ever to take. The point is that *any changes* to a game issued in a a patch which does entirely innocuous things like slightly changing the memory footprint or reordering internal shader instruction--changes which I might add are common occurances in 3d game patches which the devs make to improve the game--will break nVidia's driver detection routines and render their application-specific driver optimizations void.

I used UT2K3 as an example of how nVidia is much more interested in how 3d games can be used as benchmarks than it is in how people may simply play and enjoy 3d games. Had nVidia been interested in how UT2K3 was used as a game it never would have hard-coded its drivers to ignore any calls for full trilinear--something, I might also add, which is a distinct departure from the way in which nVidia has supported trilinear filtering historically. The only rational reason for doing something like that is that the company is more concerned with how its products run UT2K3 *as a frame-rate benchmark* than it is in how much people may play and enjoy the game without regard to it being used as a frame-rate benchmark. Considering that it is now reported that nVidia has eliminated full trilinear filtering globally in its drivers for all 3d games, it's even easier to see how benchmarks are all that concern the company at present, despite whatever chatter their PR department makes concerning the company's fondness for "real 3d games." It's quite clear to me that what nVidia is fond of are frame-rate benchmarks, and their treatment of trilinear filtering in their drivers should crystalize that concept beyond any doubt.

 

[Hanners]

I don't think you understood the point... The point is that since nVidia's "unified compiler" appears not to exist, at least as described by nVidia PR, and that what is actually occurring in the nVidia drivers is the same old application-specific optimizations nVidia's defended all year long (which depend on the ability of the drivers to detect the application being run in order to implement a specific environment strictly for that application), that whenever a game dev releases a patch which changes memory footprint or its internal shader code, nVidia's drivers will suffer exactly as they did when FM issued its patch.

But the compiler does exist - It just happens that nVidia have thrown in a shedload of application specific optimisations under the same 'compiler' banner.

 

[Nazgul]

Dagnabbit, the HardForum database is always down when I go to check out the fun in times like this

 

[Rolf N]

I have been re-hashing the issue with Kyle and someone from AVAULT.COM in this thread, if anyone wants to join the fun ..........................
http://www.hardforum.com/showthread.php?s=&threadid=688550&perpage=15&pagenumber=1

Those forums are currently undergoing a minor database implosion, it seems

 

[WaltC]

But the compiler does exist - It just happens that nVidia have thrown in a shedload of application specific optimisations under the same 'compiler' banner.

OK, fine... What I've been saying, though, is that the compiler does not exist as described by nVidia PR, which I think is evident, since the changes FM made could not have broken the compiler, as it has been described by nVidia PR. IE, the "unified compiler" is of far less importance than nVidia PR has represented. Driver-detection optimizations are still the most important segment of nVidia's "performance" plan, and it was those, instead of the compiler, that were defeated by the FM patch. But I call your attention to the fact that nVidia PR does not say this--they say that the compiler itself was broken by the patch--which is nonsense.

 

[digitalwanderer]

But I call your attention to the fact that nVidia PR does not say this--they say that the compiler itself was broken by the patch--which is nonsense.

Actually they aren't saying that anymore Walt, they've already came out and admitted they were wrong on that one and that the 340 patch does NOT disable/break the compiler.

They haven't stated where the loss is coming from if it ain't the compiler though...

 

[Nazgul]

They haven't stated where the loss is coming from if it ain't the compiler though...

Translation: The spin doctors haven't figure out how to weasel Nvidia out of this one yet

 

[karlotta]

To understand kyle, you will need a analogy... he is the George Bush jr. of the internet.

 

[PatrickL]

Can't be THAT bad

 

[digitalwanderer]

They haven't stated where the loss is coming from if it ain't the compiler though...

Translation: The spin doctors haven't figure out how to weasel Nvidia out of this one yet

I can just picture a few people in their PR dept walking around pulling their hair shouting, "You told them the truth?!?! What in the blue-bloody HELL were you thinking?!?!?" and panicking as they try and figure out the weasel angle....

 

[CorwinB]

Those forums are currently undergoing a minor database implosion, it seems

Nah, it's just the "database access compiler" technology doing its work.

 

[WaltC]

But I call your attention to the fact that nVidia PR does not say this--they say that the compiler itself was broken by the patch--which is nonsense.

Actually they aren't saying that anymore Walt, they've already came out and admitted they were wrong on that one and that the 340 patch does NOT disable/break the compiler.

They haven't stated where the loss is coming from if it ain't the compiler though...

Well, I'm basing my comments on what Perez said as I quoted earlier in the thread. Perez exists at the heart of the nVidia PR machine, whereas Luciano Alibrandi, European Product PR Manager for NVIDIA Corporation, does not... Certainly, what is coming out of nVidia Europe is a bit more intelligent, if no less evasive, than what Perez has said as I've quoted. When Perez makes a similar retraction to his own comments, then I'll be able to consider that nVidia has indeed retracted Perez's earlier statement. As it is, we've got Perez with one distinct explanation, and Alibrandi with yet another--and as you point out, Alibrandi only rules out the compiler but has "no explanation" otherwise.

When nVidia makes a "unified" PR statement to the effect that "our performance suffers because the FM patch renders our drivers unable to detect 3dmk03, and therefore makes us legal under the FM guidelines," then I'll accept that nVidia has indeed retracted its earlier statements.

But for some reason, I tend to think we aren't going to see a statement like this out of nVidia, "unified" or not, any time soon...

 

[Rolf N]

They haven't stated where the loss is coming from if it ain't the compiler though...

Translation: The spin doctors haven't figure out how to weasel Nvidia out of this one yet

I can just picture a few people in their PR dept walking around pulling their hair shouting, "You told them the truth?!?! What in the blue-bloody HELL were you thinking?!?!?" and panicking as they try and figure out the weasel angle....

Suggestion:
Constructivism (if that's the correct term anyway).

You know, it was true, but as all things in life "true" depends on perspective. What is reality? What is truth? Reality is just an illusion created by perception blah blah ...
At the end of the day, we firmly believe that Tomb Raider: Angel Of Darkness performs exceptionally well on our hardware. Thanks for listening, gob bless ya, g'night.

 

[WaltC]

Can't be THAT bad

It's much worse...he's the "Al Gore Jr." of the Internet...

 

[digitalwanderer]

Well, I'm basing my comments on what Perez said as I quoted earlier in the thread. Perez exists at the heart of the nVidia PR machine,

You forgot "dark and evil". It should be "Perez exists at the dark and evil heart of the nVidia PR machine".

whereas Luciano Alibrandi, European Product PR Manager for NVIDIA Corporation, does not... Certainly, what is coming out of nVidia Europe is a bit more intelligent, if no less evasive, than what Perez has said as I've quoted. When Perez makes a similar retraction to his own comments, then I'll be able to consider that nVidia has indeed retracted Perez's earlier statement. As it is, we've got Perez with one distinct explanation, and Alibrandi with yet another--and as you point out, Alibrandi only rules out the compiler but has "no explanation" otherwise.

Hmmm, I never thought of that. nVidia's US offices haven't even opened yet, it should get interesting in a few hours.

When nVidia makes a "unified" PR statement to the effect that "our performance suffers because the FM patch renders our drivers unable to detect 3dmk03, and therefore makes us legal under the FM guidelines," then I'll accept that nVidia has indeed retracted its earlier statements.

When nVidia makes a statement like that I'll video tape myself doing my silly "geek-happy dance" and host it on the web for all to laugh at! (Read: it ain't gonna happen. )

If nothing else, it looks like it'll be an interesting Friday. 8)