PDA

View Full Version : nVidia: "...and that is in fact not true,”


digitalwanderer
11-14-03, 09:53 AM
Link to quote on X-Bit Labs (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.”

The Baron
11-14-03, 10:18 AM
but wait. I'm confused.

they're now denying that 3DMark affects the compiler... so what kills performance? no explanation, but also no FUD anymore. could this in fact be the moment prophesied by many where NVIDIA begins to come clean?

digitalwanderer
11-14-03, 10:20 AM
Originally posted by The Baron
but wait. I'm confused.

they're now denying that 3DMark affects the compiler... so what kills performance? no explanation, but also no FUD anymore. could this in fact be the moment prophesied by many where NVIDIA begins to come clean?
Yeah, I know....things are getting so bad for them they're actually resorting to the truth! :eek:

Hold me, I'm scared! :eek:

The Baron
11-14-03, 10:23 AM
Originally posted by digitalwanderer
Yeah, I know....things are getting so bad for them they're actually resorting to the truth! :eek:

Hold me, I'm scared! :eek:
If it is true--did I call it or what? I said a few months back, "NVIDIA will come clean with driver cheats before the NV40 launch so that it can regain much of its lost respect in the enthusiast community and therefore have a better-received launch." Okay, so I expected them to do this two or three months ago, but that was when NV40 was still coming out this month.

But jeez. Gives me hope that we will in fact see a control panel option to enable or disable application-specific optimizations.

digitalwanderer
11-14-03, 10:32 AM
Originally posted by The Baron
If it is true--did I call it or what? I said a few months back, "NVIDIA will come clean with driver cheats before the NV40 launch so that it can regain much of its lost respect in the enthusiast community and therefore have a better-received launch." Okay, so I expected them to do this two or three months ago, but that was when NV40 was still coming out this month.
I don't think this was a planned PR maneuver at all, I think this was just a re-grouping from the PR debacle that was yesterday for them and they realized they had said something which was not only untrue but has already been disproved so they HAD to issue a retraction just to cover their arse legally. (FM could have nailed them for slander)

It's nice to see a bit of honesty, but they only resorted to it 'cause their backs are flat-up against a wall.

But jeez. Gives me hope that we will in fact see a control panel option to enable or disable application-specific optimizations.
Keep dreaming, they'd have to admit that they have app-specific opts which they STILL haven't done. (They now officially have no reason for their score drop. :) )

The Baron
11-14-03, 10:37 AM
You think PR guys are stupid; I think they're geniuses. Either way--Futuremark isn't suing anyone, I would imagine, since NVIDIA would probably just buy them if they brought a suit. NV could have kept the compiler claim going to keep the fanboys happy and turn their rage against Futuremark, but they didn't. THAT is a DEFINITE change for them.

Think about it--over the past several months, NVIDIA hasn't seemed to give a damn what the people who understood the subjects thought; it's all about the OEMs, the fanboys, and the Average Joes. The ONLY people who realized that the compiler claim was complete bull was the people who knew what they were talking about. This is not business as usual for NVIDIA at all.

digitalwanderer
11-14-03, 10:54 AM
I don't know, but I do have one bit of knowledge that's bugging me and doesn't fit but would fit just peachy with your "super-genius PR" theory.

nVidia is a beta member of FM, right? So they received the 333 patch (same as 340 patch, just it's beta designation) in advance of FM releasing it just like B3D and some other beta partners did....so they KNEW it was coming too, yet they reacted totally confused and gave out mixed signals yesterday?

That does NOT sound like the nVidia PR dept I've come to know and fear, it seems like an awfully big "ooops!" to not know that the patch was coming and going to smack 'em. I see it as one of two things:

1. Communication is so bad that PR didn't know the patch was coming out or didn't know it was going to shoot down their performance.

2. It was a carefully orchestrated plan to make it look like they were panicking, caught themselves in a lie, and started this morning down the golden path of righteousness to rid their company of the evil nasitiness that has infested them and get back to customer satisfaction as their main goal. (Your "genius PR" theory)

I just don't see them being clever enough for #2, I really don't.

The Baron
11-14-03, 10:55 AM
Oh, I do... I mean, look at how much effort they've put into Futuremark over the past six months. They wouldn't screw this up.

Reaper106
11-14-03, 10:59 AM
Originally posted by digitalwanderer
Yeah, I know....things are getting so bad for them they're actually resorting to the truth! :eek:

Hold me, I'm scared! :eek:

There there Dig, It's gonna be alright.:afraid:

digitalwanderer
11-14-03, 11:10 AM
http://www.nvnews.net/vbulletin/attachment.php?s=&postid=232211

hohum
11-14-03, 11:45 AM
Well to be honest you could see straight away that 'disabled our compiler' etc was a load of rubbish. Example: noticeable speed drops in the game tests with the new patch, yet the ps 2.0 test (the two elephants) is exactly the same as before - no performance decrease. Now surely if their compiler had been disabled, there would be a drop in this test too? After all this particular area was exactly the reason for creating the compiler - poor ps 2.0 performance...

Reaper106
11-14-03, 12:12 PM
Originally posted by digitalwanderer
http://www.nvnews.net/vbulletin/attachment.php?s=&postid=232211

ROFLMFAO, Damn Dig I can always count on you.:firedevil

volt
11-14-03, 01:29 PM
Anyway, how their Unified Compiler can get disabled? It can't, simple as that.

sxotty
11-14-03, 01:42 PM
Heh I saw that on bluesnews earlier dig...

The way I see it is

the PR department called the cheats part of the compiler, then they realized hmmm this will besmirch the good name of our much touted compiler and decided to call the compiler the compiler and pretend the cheats don't exist.

Uttar
11-14-03, 03:45 PM
Originally posted by sxotty
Heh I saw that on bluesnews earlier dig...

The way I see it is

the PR department called the cheats part of the compiler, then they realized hmmm this will besmirch the good name of our much touted compiler and decided to call the compiler the compiler and pretend the cheats don't exist.

That's what I've been posting on forums for the last 2 days :P

Dig, read your PMs. A rather strange but unlikely theory of why they operated that way is included. FYI only.


Uttar

StealthHawk
11-14-03, 04:03 PM
Originally posted by The Baron
You think PR guys are stupid; I think they're geniuses. Either way--Futuremark isn't suing anyone, I would imagine, since NVIDIA would probably just buy them if they brought a suit. NV could have kept the compiler claim going to keep the fanboys happy and turn their rage against Futuremark, but they didn't. THAT is a DEFINITE change for them.

They may be geniuses at manipulating and confusing the sheep, but to anyone with a slight bit of intelligence PR is nothing more than a few people trying to pull a veil over your eyes very discretely.

They contradict themselves and others all the time.

The compiler claim was complete rubbish to begin with. The interesting thing now is that NVIDIA is not offering up an excuse for lower performance.

If it is true--did I call it or what? I said a few months back, "NVIDIA will come clean with driver cheats before the NV40 launch so that it can regain much of its lost respect in the enthusiast community and therefore have a better-received launch." Okay, so I expected them to do this two or three months ago, but that was when NV40 was still coming out this month.

Just like last time, I say it is too early to jump to any conclusions. Remember the Optimization Guidelines and what a disappointment those turned out to be?

WarheadMM
11-18-03, 12:25 PM
The reason why Nvidia lost soo many points on 3dmark03 is because they didnt make a unfied complier for the vertex shader yet. So what nvidia did is they tweak the vertex shader to a massive point. When nvidia releases newer drviers it will have the unifided compiler for the vertex shader

saturnotaku
11-18-03, 02:16 PM
Originally posted by WarheadMM
When nvidia releases newer drviers it will have the unifided compiler for the vertex shader

Couple questions:

1) What's your source on this?

2) Would said compiler for the vertex shader be enough to a) be a legitimate optimization and b) boost scores back to their "pre-cheat" levels?

Sazar
11-18-03, 02:55 PM
seconded...

this is the first I heard of that theory warhead... though it is possibly feasible :)

but its as feasible as just about anything else out there right now :D

The Baron
11-18-03, 03:43 PM
See, it's not true. Because, if it were true, it would have been included. And vertex shaders aren't that performance intensive compared to vertex shaders.

How would improving vertex shaders ten or fifteen percent make up for a ten or fifteen percent total performance drop in 3DMark03? Answer? It can't. Why? Because they had clip planes and shader replacement before.