Leaked ATi internal slides

I'm just curious, how would they know a "competitor" is distributing the slide? Would be better to just comment on the slide or "provide" the rest of the slides if they want to clear the air. To say, "A competitor..." sounds like a weak PR move. I trust PR people as much as I can throw a car.

 

Ah, but what if it was one of the smaller players, knowing that everyone would assume it was Nvidia, thus adding to the chance they'll go down in flames and the real leaker could move up in the world... 8)

 

What would a smaller player gain? Which smaller player could come in and pick up sales because of discrediting ATI or Nvidia? I can't think of anyone who has the hardware. Heck, even Nvidia doesn't have the hardware, they are just hanging on with brand recognition and marketing.

 

Well, I was only joking... I was thinking about XGI or S3 trying to clear the path for a rise to prominence

 

These slides are absolutely GREAT and dictate EXACTLY how an IHV should morally proceed with touting and marketing their products. I find it laughable how anyone (especially if the source is NVIDIA) could try and twist these as being some sort of negative. They are, in fact, the complete opposite.

The line "Influence Benchmarks in our Favor" is EXACTLY how an IHV should morally approach performance measurement benchmarks! Notice it says "Influence BENCHMARKS" and not "Influence our Drivers or Products to FAKE" etc.etc.

To "Influence Benchmarks in our Favor" stipulates that through persuasion try and convince benchmark providers to use features or other uses that will provide the best benefit to ATI class hardware. This is the proper way to market a product.

A good example- if NVIDIA wanted to get good scores, they could try to persuade a benchmark company to use lots and lots of stencil shadows and framebuffer reads in order to give a larger impact to their hardware over ATI's. NVIDIA has done this for YEARS with Futuremark by insisting previous versions use the smallest, least thrashable sized textures with unheard of 2-vertex T&L ganging for all geometry through static paths. This is why 3dmark2000-> had such a HUGE lead for NVIDIA Products. It wasnt fake or unreal performance, it was just insanely incorrect when compared to videogames since it would take 14+ years to write a dynamic game engine with this degree of specialization.

Of course, now consumers of 3d accelerators have become a bit more knowledgable and saavy so something this extreme would clearly not be possible, but similar efforts to a lesser effect could be approached. You could basically make two types of benchmarks to sway results towards one IHV or another- one being a more Doom3-ish benchmark (low precision shaders, lots of stencil,etc.etc.) or a heavy DX9 style benchmark (lots of higher precision shaders, larger textures, etc.etc.)

The leading IHV's both have their strengths and weaknesses and you could pretty much set-out to show massive superiority for one versus the other, and do so without any real fear of recoil. After all- Doom3 is going to be a stencil-heavy, shader-light game, so you could just point an NVIDIA-centric benchmark at Doom3 as the model and be kosher for "real gaming" use comparison/accuracy. ATI has the same effect with it's emphasis on high-precision shader performance, so you could point any ATI-centric benchmark at every other DX9 game and benchmark so far as being the model.

 

Damn DW, you're such a muckraker!

At least you can't be accused of going just after Nvidia now eh?

 

VERY VERY INTERESTING.

Seems elitebastard had a 3rd leaked slide which didn't quite get the same airing hmm.

http://www.elitebastards.com/pic.php?picid=forum/album_pic.php?pic_id=93

 

AHAHAH YOU BASTARD YOU DELETED IT THAT LINK WAS CORRECT. :/ damn I'm gonna search through the inet cache now!

The html even still has the correct image size but thanks to no cache I dont got the picture.

The image was 1456x1104 its nice to see that the images either side of it still work.

The slide was comparing ATI Gear the game or whatever to Nvidia the way it ment to be played.

Now I won't mention what it said more closely I'll let DW explain if he wants

Nah lets bust him he tried ot make a fool of me

FRANKLY DW has bashed ATI
the slide that he forgot to show contained a comparision between the two such points as

ATI

Encourage developers to use standard apis

NV

Encourage developers to use CG locking them into it and similar to what happend with glide.

 

THANK YOU Bolloxoid

Someone that agrees with my POV.

Maybe we can also tell DW that Ati is a corporation and are out to make money, just like every other corporation on the planet?

Any corporation would be stupid NOT to influence things to their favour.

 

Bah... hell... so those slides are the real deal... They still look odd to me but what do I know.

Well at least we have more drama... It seems the majority of the community was smart enough to not take the bait on that one and call it like it is. Impressive to say the least.

Wating for the next episode of this soap opera.

 

I don't want to get caught in a Russ-like situation (To pull a Russ; To Russ), but the point obviously needs to be made:

While "benchmarks are made to be neutral," ATi is not.

Remember FM wants nV in the fold to gain insight and tips into their future architecture. Consider this a form of influencing 3DM04 in nV's favor.

FYI (OK, to avoid being painted with the Russ brush ), I shot this off to happy fun Fuad:

 

Solomon, you're dead right with your take on PR people. Having said that at least in Yahoo Finance someone's saying that Brian Burke has distributed the ATI slide. I wonder whether he's given any accompanying remarks while doing it...

http://finance.messages.yahoo.com/bbs?.mm=FN&action=m&board=15969433&tid=atyt&sid=15969433&mid=76122

(and yes, if digi really posted that I'm inclined to agree)

- Tom

 

When I saw the slides I wondered how out of context they were and how inflammatory the first sentence would be seen as. Influence in our favour is too open to be criticsed beyond it being too open! You could infer it means any or all of:

1) Follow OPEN standards more
2) Use more advanced features
3) Use more advanced shaders
4) Use shaders structured more to ATi's capabilities
5) All the NVidia types of tricks and cheats (IQ loss, optimisations)

I love slide 3 Frankly I see this as a lame attempt by a used to be great company to stall until they have their act together and overall it is hurting the industry and its consumers. Someone should really read them the riot act.

 

Errr...I took it to go and convince benchmark writers to test things the ATI parts were good at (or opponents parts were bad at) and avoid doing the polar opposite.

At least that's what I would do if I were trying to sell graphics chips.