NVIDIA's offical response to Futuremark

[DivotMaker]

Originally posted by SurfMonkey
From Catalyst Maker over @ Rage3D...

I don't think that's a very good response, they're basically saying we endorse NVidia's method too and just to prove we are no better than they are we will devote time and money into ensuring the consumer and the industry both suffer.

I think if you put yourself in Terry's shoes, you would likely be as frustrated and angry as he is. I can understand how he feels and i think if you asked him today, he would temper his remarks somewhat and respond a tad less emotionally. I don't know him well, but he seems like a tremendous person and a great asset to ATI and the ATI community.

Keep your chin up Terry!

[R.Carter]

Originally posted by Morrow
What do you prefer: A benchmark which indicates how your graphiccard will perform for example in Doom3 or HL2 or a benchmark which indicates how your graphiccard will never perform since the bench does makes use of an graphic engine which is not feasible enough to be used anytime in real game?

Heh! If someone really cared about Doom III or HalfLife 2 or other "real world" applications then I would think it's best to test using those specific applications, right?

If you only care about how good the architecture itself is, then I'd use an application that hasn't been optimized. Of course, that isn't representative of "real world", but then who cares?

Not everything is going to be optimized and if the architecture is THAT sensitive (24%) to lack of optimizations then people might be interested in know that fact before they buy.

[UDawg]

Originally posted by Ady
If you don't understand it then go away.

It's his and many others choice to boycott a company due to bad ethics or any other reason. Ones choice is not yours to judge.

Go away? Nice thing to say, compared to someone who wants a whole company do go belly up. This is why I do stay away because of flaming children like your self.

Ones choice is not yours to judge.

I just did judge it.

BTW boycotts for the most part don't work. Good luck.

[Zenikase]

Originally posted by Neural
Forgot something, I agree with most of what you say about the CPU/GPU achitectures and drivers, but i dont know if microsoft develops the drivers for different CPUs, I could be wrong but I would have thought that their makers made the drivers, submit for WHQL, and are then implemented in the OS. Not all CPUs are natively supported by a standard driver, I had to get a driver update from windows update for my barton in order to take advantage of the extra L2 cache. Windows update also offers some older GPU drivers, but they dont try to keep up with the ever changing GPU drivers.

If there are such driver updates for Windows, then they are simply for cases like that where the features of newer CPUs are not recognized and enabled natively by the OS. I don't know who creates these drivers either, but there's no way a CPU manufacturer can cheat in the way video card companies do, since there's no compiler present in the driver. The CPU will do its work to the best of its ability based on the application's code, the compiler that was used to make that application, and the computational efficiency of the CPU.

[Quinn1981]

Originally posted by Solomon
Apparently you never been to court before. I can tell you straight out it's really not a "legal" thing but, "How much money can you bring to the table" They will drag it out as long as possible. It's an endurance thing.. It sucks...

Regards,
D. Solomon Jr.
*********.com

I think it's more about how far your pricey lawyers are willing to go. I know about throwing money around and dragging crap out, but Futuremark also has friends that might help them out and could have valid points that could speed up a trial and shut nVidia's people up. It's not just a money game in court from my view. There are a lot of companies that would like to keep Futuremark around. There's a lot of interest that equals money with Futuremark. I think that's why we are seeing a little settlement like this. They know they are better off making themselves look bad for now so they can do better in the future. It would be silly for Futuremark to say all those things in the PDF if they didn't know it could lead to nVidia coming to the offering table. Or at least I think so.

[Morrow]

Originally posted by Solomon
WoW! Apparently you haven't heard of the "Clip Plane" issue in 3DMark03 with the 44.03 drivers huh? Hehehe.

nope, I didn't miss both the extremetech and the FM audit report but unfortunately most people here seem to miss the big picture about this joint statement and what it means for the future:

When 3dmark03 was released, nvidia had quit the beta member team because they did not agree with FM's philosophy about benchmarking "standard performance". Nvidia knows that (IQ loss-less) optimizations are an important factor for demonstrating the full power of GeForce cards. All important games (especially ID games) make full use out of those optimizations but now comes FM and released a "gaming benchmark" as they call it but only relies on standard DX code which is never used that way in games... of course nvidia quit

Then came FM's statement a few days after the release of 3dmark03 that ANY change in the driver code to increase performance in 3dmark03 is not legitimate. Nvidia again couldn't understand how FM wants 3dmark03 to be the benchmark for future game performance when they code their engine exactly the way future real game engines will never be coded. So nvidia decided to boycott FM by saying that 3dmark03 is not reliable because of easy to make driver optimizations to synthetically increase the 3dmark03 score and they even went as far as demonstrating it in their 44.03 drivers.

But this whole optimization/cheat disbute now got finally a postive turn in a way that allows nvidia and all other graphiccard manufactures to include driver optimization (like done in real games) in future versions of the popular 3dmark benchmark (not yet including 3dmark03) making this bench or more real-life indicator that it currently is.

This means that nvidia doesn't need to hide anymore optimizations in their drivers, no longer needs to replace shaders effecting IQ but can go straight to FM and help them on how to accomplish much faster code with the same IQ. It will be all more easy and transparent for everyone. It's a clearly win/win situation.

[Solomon]

Originally posted by Morrow
nope, I didn't miss both the extremetech and the FM audit report but unfortunately most people here seem to miss the big picture about this joint statement and what it means for the future:

When 3dmark03 was released, nvidia had quit the beta member team because they did not agree with FM's philosophy about benchmarking "standard performance". Nvidia knows that (IQ loss-less) optimizations are an important factor for demonstrating the full power of GeForce cards. All important games (especially ID games) make full use out of those optimizations but now comes FM and released a "gaming benchmark" as they call it but only relies on standard DX code which is never used that way in games... of course nvidia quit

Then came FM's statement a few days after the release of 3dmark03 that ANY change in the driver code to increase performance in 3dmark03 is not legitimate. Nvidia again couldn't understand how FM wants 3dmark03 to be the benchmark for future game performance when they code their engine exactly the way future real game engines will never be coded. So nvidia decided to boycott FM by saying that 3dmark03 is not reliable because of easy to make driver optimizations to synthetically increase the 3dmark03 score and they even went as far as demonstrating it in their 44.03 drivers.

But this whole optimization/cheat disbute now got finally a postive turn in a way that allows nvidia and all other graphiccard manufactures to include driver optimization (like done in real games) in future versions of the popular 3dmark benchmark (not yet including 3dmark03) making this bench or more real-life indicator that it currently is.

This means that nvidia doesn't need to hide anymore optimizations in their drivers, no longer needs to replace shaders effecting IQ but can go straight to FM and help them on how to accomplish much faster code with the same IQ. It will be all more easy and transparent for everyone. It's a clearly win/win situation.

Um... The clip plane is not an optimization! As soon as you can see that the sooner you will realize that nVidia cheated and is buying it's way out of the paper bag.

Regards,
D. Solomon Jr.
*********.com

[Solomon]

Originally posted by Morrow
When 3dmark03 was released, nvidia had quit the beta member team because they did not agree with FM's philosophy about benchmarking "standard performance".

So from this you are saying you rather have an nVidia optimized benchmark then a standard benchmark? There is a reason why it's a standard benchmark. nVidia can optimize for the benchmark. Hell no one is saying they can't. They can't cheat is what it's all about. The "Cheating Drivers Hidden Dragon" picture says it all. The clip plane says it all.

From what I'm getting from you is that it's o.k. to have all that not show and still be legit? You got to be kidding me.

Regards,
D. Solomon Jr.
*********.com

[jAkUp]

Originally posted by BigBerthaEA
I am not sure where you are getting your information, but EA Sports dominates the top 5 positions in PC Sports game sales. They also dominate the console sports sales charts as well. Now, I will grant you that PC sales are not nearly as good as console sales, but I think this is primarily true throughout the industry with a few exceptions.

EA nor EA Sports are going anywhere but up. The last fiscal quarter, EA had it's first ever $ Billion revenue quarter. Largely fueled by Madden 2003, The Sims, BF 1942, MOHAA, etc.

im not saying their revenue is going down... ive been an ea sports fan ever since 98, and madden, live, nhl, triple play 98... 99 was great was great too... 2001 was a big turning point for the company imo... all their sport titles turned upside down... left alot of sports fans upset. madden, nba, nhl, and triple play 2001 we're pretty dissappointing for the strides the earlier games made... the years after that we're only mild steps up. older years the difference between say 98>99 was night and day... the triple play series took a year off cuz it mad sucked... and needed an overhaul... (not sure about their new game)

ea used to be my favorite company... very nice games imo... and need for speed... my favorite racing game, was recently ruined on the ps2 and pc flamming barrels??!!! gimme a ****in break!!! lmao

[Zenikase]

Originally posted by Morrow
nope, I didn't miss both the extremetech and the FM audit report but unfortunately most people here seem to miss the big picture about this joint statement and what it means for the future:

When 3dmark03 was released, nvidia had quit the beta member team because they did not agree with FM's philosophy about benchmarking "standard performance". Nvidia knows that (IQ loss-less) optimizations are an important factor for demonstrating the full power of GeForce cards. All important games (especially ID games) make full use out of those optimizations but now comes FM and released a "gaming benchmark" as they call it but only relies on standard DX code which is never used that way in games... of course nvidia quit

Then came FM's statement a few days after the release of 3dmark03 that ANY change in the driver code to increase performance in 3dmark03 is not legitimate. Nvidia again couldn't understand how FM wants 3dmark03 to be the benchmark for future game performance when they code their engine exactly the way future real game engines will never be coded. So nvidia decided to boycott FM by saying that 3dmark03 is not reliable because of easy to make driver optimizations to synthetically increase the 3dmark03 score and they even went as far as demonstrating it in their 44.03 drivers.

But this whole optimization/cheat disbute now got finally a postive turn in a way that allows nvidia and all other graphiccard manufactures to include driver optimization (like done in real games) in future versions of the popular 3dmark benchmark (not yet including 3dmark03) making this bench or more real-life indicator that it currently is.

This means that nvidia doesn't need to hide anymore optimizations in their drivers, no longer needs to replace shaders effecting IQ but can go straight to FM and help them on how to accomplish much faster code with the same IQ. It will be all more easy and transparent for everyone. It's a clearly win/win situation.

The problem is that it's not the same IQ. Take a look at some of the infamous screenshots of the scenery that's hidden in the regular benchmark. There are multiple major inaccuracies and errors in how the tests are rendered, due to nVidia's drivers intentionally replacing Futuremark's shaders with their own proprietary shaders, and inserting arbitrary clip planes in areas that are outside the regular viewspace. This is not an optimization; this is a fast and cheap method of getting higher scores by ruining the original intended result of the 3DMark03 team. It also destroys the possibility of creating comparable results.

When a driver detects an application and/or certain parts within the application and modifies code to its own liking, it is outright cheating.

[gordon151]

Originally posted by Morrow
nope, I didn't miss both the extremetech and the FM audit report but unfortunately most people here seem to miss the big picture about this joint statement and what it means for the future:

When 3dmark03 was released, nvidia had quit the beta member team because they did not agree with FM's philosophy about benchmarking "standard performance". Nvidia knows that (IQ loss-less) optimizations are an important factor for demonstrating the full power of GeForce cards. All important games (especially ID games) make full use out of those optimizations but now comes FM and released a "gaming benchmark" as they call it but only relies on standard DX code which is never used that way in games... of course nvidia quit

Then came FM's statement a few days after the release of 3dmark03 that ANY change in the driver code to increase performance in 3dmark03 is not legitimate. Nvidia again couldn't understand how FM wants 3dmark03 to be the benchmark for future game performance when they code their engine exactly the way future real game engines will never be coded. So nvidia decided to boycott FM by saying that 3dmark03 is not reliable because of easy to make driver optimizations to synthetically increase the 3dmark03 score and they even went as far as demonstrating it in their 44.03 drivers.

But this whole optimization/cheat disbute now got finally a postive turn in a way that allows nvidia and all other graphiccard manufactures to include driver optimization (like done in real games) in future versions of the popular 3dmark benchmark (not yet including 3dmark03) making this bench or more real-life indicator that it currently is.

This means that nvidia doesn't need to hide anymore optimizations in their drivers, no longer needs to replace shaders effecting IQ but can go straight to FM and help them on how to accomplish much faster code with the same IQ. It will be all more easy and transparent for everyone. It's a clearly win/win situation.

Why do you keep comparing 3DMark03, which is based on a standardized format with no vendor specific extensions to an OpenGL format which uses vendor specific extensions? Why do you expect 3DMark03 to be a vision of Doom III performance which by Carmacks own accord, doesn't make full use of all the new features in the current DX9 hardware. It's just a case that Doom III requires a DX9 card purely on the merits that it just needs the raw processing power to calculate all the shadowing effects and complex textures.

Furthurmore ATI was the only company of the two that used an optimization in 3DMark03 that had no effect on IQ at all since it was just an optimization of an existing routine. Two highly influential developers have gone on the record as saying that such a thing is an extremely valid optimization. What they didn't say was a valid optimization was anything that implemented a loss of visual quality from the intended output and in the case of synthetic benchmarks anything that reduces the workload of one specific card.

It would also be nice to point out that benchmarks are and should be solely a tool for users. How can one accurately be sure of any benchmarking tool where the manufacturer of said product being benchmarked has an active hand in exactly what is being done in the program to gauge his card? Before this high level members had an influence of "suggesting" what was going into the benchmark, but now it looks like they (or more specifically Nvidia and not the users) have a more direct hand in determining what is fair or not towards their product.

I for one can't trust any timedemo from now on from both ATI and Nvidia (more so for Nvidia absolutely) and subsequently will be more inclined to make buying decisions from actual benchmarks of playing performance like that which were done in nV News. Nvidias actions have so plainly caused such a scenario to be the case for quite a lot of people.

[digitalwanderer]

Originally posted by Solomon
Um... The clip plane is not an optimization! As soon as you can see that the sooner you will realize that nVidia cheated and is buying it's way out of the paper bag.

I wouldn't even waste your time arguing with him. Either he's a totally brainwashed fool who actually believes nVidia's version of reality or he's just looking to provoke a reaction. Whichever it is, I'm a little tired of wasting me time on the clueless and the attention seekers.

This whole thing just really sucks and makes me feel all dirty about something I love, very unhip.