NVIDIA Corporation has demonstrated a document criticizing its rival’s latest products as well as revealing certain unpleasant facts about ATI Technologies to press members as well as its own managers.
The History Repeating
The presentation that is marked as NVIDIA’s internal document, says that ATI’s latest RADEON X800 PRO and RADEON X800 XT graphics processors are based on the last year’s architecture and have very suspicious drivers in terms of possible cheating and unfair optimizations. The document also claims that ATI Technologies could not implement Shader Model 3.0 in the RADEON X800 series and also mislead reviewers and customers about core-clocks of the flagship offering RADEON X800 XT.
NVIDIA’s representatives confirmed [H]ard|OCP web-site the authenticity of the slides from the presentation published by 3DCenter.de web-site.
NVIDIA is pretty well-known for pass negative judgments on competing products. In 2001 the company issued a special document for its sales managers and some clients telling that PowerVR’s KYRO graphics processor does not worth attention and is not a potentially feasible technology.
PowerVR’s KYRO competed with NVIDIA’s GeForce2 MX400 at about $150 price-point, showing decent performance in almost all games of that time, but did not support certain features, such as T&L engine. In a lot of cased the KYRO was faster compared to more expensive NVIDIA GeForce2 GTS and ATI RADEON 256 64MB DDR products.
In 2003 NVIDIA also released a document claiming that Futuremark’s 3DMark03 benchmark was unfair and provided certain benefits to arch-rival’s ATI RADEON graphics processors.
ATI’s flagship offering RADEON X800 XT graphics card outperformed NVIDIA’s top-of-the-range GeForce 6800 Ultra in a lot of cases based on the result of numerous gaming benchmarks. But ATI’s RADEON X800-series does not sport some capabilities NVIDIA’s new chips are able to execute, such as Shader Model 3.0.
Belittling Rivals’ Instead of Praising Own Products
Earlier this month ATI Technologies supplied reviewers of ATI RADEON X800-series of graphics cards a document suggesting that they should disable certain trilinear filtering optimizations of NVIDIA GeForce 6800 Ultra GPUs from the ForceWare drivers to ensure “fair competition”. Later it was discovered that ATI also had similar optimizations of trilinear filtering for its RADEON X800-series hardware that could not be disabled by the drivers.
While this is not news that hardware developers bash each other internally and sometimes even issue special documents to humiliate rivals in the eyes of potential clients, the leak of this kind of documents is not praised by the community of hardware enthusiasts, who often start to criticize authors of the documents and issue negative feedback.
Both leading makers of central processing units – Intel and AMD – were noticed spreading documents humbling each other too.
Comments currently: 17
Discussion started: 05/26/04 10:36:56 AM
Latest comment: 05/29/04 03:50:00 PM
Expand all threads |
Collapse all threads
[1-6]
1.
This is a pretty unfair assessment of nVidia, as ATI has done very similar, like in their "Save the Nanosecond" presentation. You should have mentioned such things. It's also very unfotunate that this article makes the claim that "RADEON X800 XT graphics card outperformed NVIDIA’s top-of-the-range GeForce 6800 Ultra in a lot of cases based on the result of numerous gaming benchmarks" and not explain that ATI was using an optimization until later in the article. Ultimately, this is a bash of nVidia, and is hypocritical to say this.
[Posted by: somebody | Date: 05/26/04 10:36:56 AM]
+ expand thread (10 answers)
- collapse thread
X-bit labs published NVIDIA GeForce 6800 Ultra scores with and without optimized trilinear filtering. RADEON X800 XT managed to outperform the GeForce 6800 Ultra with the optimizations enabled.
Please check reviews by X-bit labs. We typically comment based on the results achieved by our own lab in the largest number of gaming tests possible.
[Posted by: Anton | Date: 05/26/04 12:16:39 PM]
Well, first of all, that review has some serious problems. Your Call of Duty numbers for the 6800 in the official review does not match the numbers for Tom's Hardware:
http://www20.tomshardware.com/graphic/20040504/ati-x800-1 .html
nor your own analysis of AF and AA on CoD (meaning the numbers you were getting for no AA&AF mode vs. 4xAA/16xAF simply were not support by the following):
http://www.xbitlabs.com/articles/video/display/r420_16.html
http://www.xbitlabs.com/articles/video/display/r420_21.html
Also, if the following is accurate, then your claim of X800XT beating the 6800U in a large number of games (w/o AA or AF) is simply false:
http://www.nvnews.net/vbulletin/showthread.php?t=28520
Last, turning on 6800U optimizations is the not the same as turning off x800xt's ones off. Simply put, in light of problematic numbers and counter claims of that original review, my post stands. Not to mention that you have not pointed out ATI's actions of the same thing, which is irrelavent of how the performance numbers come out.
[Posted by: somebody | Date: 05/26/04 02:47:26 PM]
We do not comment on the results achieved by other web-sites.
[Posted by: Anton | Date: 05/27/04 02:40:41 AM]
But your OWN numbers are BS. Will you comment on that?
[Posted by: somebody | Date: 05/27/04 10:38:52 AM]
Our own numbers are absolutely correct for the driver versions we used, test-system we used and graphics cards we used.
[Posted by: Anton | Date: 05/27/04 11:19:43 AM]
Then you must have screwed up somewhere between the drivers, test-system, and graphics card as you have numbers not supported by anyone else are not self-consistent either.
[Posted by: somebody | Date: 05/27/04 01:17:46 PM]
What's your intention with calling Xbit Lab's results BS? It appears that Xbit labs tend to be more neutral than most other sites, give them a break.
Call of Duty benchmark scores vary depending on what is actually used to measure the performance. Toms hardware used their own recorded demo. Do you honestly expect all others to use the Toms Hardware's recording?
The table provided by Nvidia fansite. How nice? Many sites listed there are pro-Nvidia, would you consider that valid data? I don' even see Hard OCP results there.
I'm surprised how effective Nvidia PR is in deceiving many users. The whole debate over ATI's optimization is fundamentally different from Nvidia's cheating.
Also, why is that Nvidia bashing ATI if they are so confident in the performance of their new product? I really wonder why?
[Posted by: conix67 | Date: 05/27/04 08:02:10 PM]
Obviously you have no clue what you are talking about. Read the comments of the x800 review. A number people have called it a very biased review. If you don't anything from a "pro-nvidia" site, then go check those places out yourself. If you knew anything about how HardOCP benchmarks, you would know why it would not be there.
Here we have another ATI saying "optimization != cheat" even if used unfairly. Since most review sites only show a few percent victory by ATI, it makes all the different. If nVidia were to release a similar thing in there next driver release and everyone uses it then you better not complain about it.
[Posted by: somebody | Date: 05/28/04 07:11:20 AM]
A number of Pro-NVs calling it biased. Why do they feel that way? I don't see why. What they don't seem to know is that optimzations are everywhere in Nvidia products since NV30 (some called them ch**ts). What makes you confident in saying that Nvidia doesn't have any optimziations in review drivers? How do these guys improve performance in every new driver release? Some magic? No. It's called O-p-t-i-m-i-z-a-t-i-o-n !!
Did you see the sites reported they received NV40 Ultra Extereme and GT boards and new drivers just one or 2 days before X800 NDA was lifted? Some review sites actually included Ultra Extreme scores in the X800 review benchmarks. Now, Nvidai is saying there's no such product planned. What kind of joke is this?
The real reason people are picking on ATI with their trilinear optimization was because ATI's been claiming "true" trilinear all along. What they provide is indeed "true" trilinear quality but simply not using trilinear filtering all the time. The same cannot be said for Nvidia. I wonder why only NV40's filtering seem to be vastly different from reference reders which render very closely to ATI's and Nv GF4s.
Hard OCP's results make it only easier to compare. They do use different method from other sites, but they have good reason behind it. Their method is actually quite good for people who wants to compare how their favorite games will perform on the hardware they wish to purchase, which is very informative. Remember, Hard OCP was truely biased toward Nvidia not even one year ago.
That said there's no denying NV40 appears to be a fine piece of hardware. The funny thing is whatever they have done over the last couple of years seem to be backfiring on them.
[Posted by: conix67 | Date: 05/28/04 07:43:48 PM]
Wasn't the 6800 Ultra Extreme an overclocked 6800 Ultra? If so, I think it got renamed to 6850 Ultra.
My next card will be nVidia, even with ATI top-end performance, and mid-end price/performance being better, because of better Linux support. (No, it's not going in my current box - the TNT2 M64 32MB PCI is working JUST FINE in that, thank you very much)
I'm not a nV fanboy, if ATI's Linux support gets as good as nV's, or the DeltaChrome is good and handles Linux well (from what I've heard, the UniChromes (ProSavages) are better than nV under Linux, so maybe we'll get lucky), I'll switch in a heartbeat.
[Posted by: bhtooefr | Date: 05/29/04 03:50:00 PM]
2.
The only constant in the IT Universe, are companies putting other's products down rather than promoting their own. Intel, Microsoft and nVidia are all culprits at spreadding the FUD game.
Now, FUD can be true and it can be a pack of lies (I'm not going to comment on which the ATI cheating aligations are, as I've personally seen no proof other than forum postings), but IMHO it's just a sign of unimaginative marketing. It's just a shame that it works so well.
Anyway, about this supposed cheating thing, they've got room to talk!! It seems a bit hypocritical to me.
srg
[Posted by: srg86 | Date: 05/26/04 11:06:41 AM]
3.
What I don't get with all these people that stand by nVidia, is how do any filtering algorithms ATi uses invalidate their comparisons? With ATi's algorithm they are still able to deliver the same or better IQ - which is what determines whether it is a cheat or a legitimate optimization.
What I don't get is that reviewers let these fans of companies get in the way of making that distinction - or compromise with competetitors and call something foul when up untill now their IQ recieved praise.
[Posted by: tfranzese | Date: 05/26/04 03:29:19 PM]
+ expand thread (1 answer)
- collapse thread
That is why we included two types of NV40 scores in the latest "Clash of the Titans" article.
[Posted by: Anton | Date: 05/27/04 04:00:02 AM]
4.
I've chosen Nvidia several times but not due to the optimization issue. I prefer the more advanced shader technology because with that I don't feel I'll have to change out cards when games start popping up using the technology. ATI wants to charge you twice to get the perf now and 3.0 later. I've had a lot of personal experience with the drivers of both, using a 9800 256mb for the ATI side, and the drivers are still dramatically superior in the NV camp. That's subject to some degree of preference, but a lot of new games still consistently create bugs in ATI implementations where NV runs fine with the game right out of the box. I guess lastly, I'm not all that worried if I lose a few fps for better quality, better drivers and better technology.
I very much realise that this is not the same choice everyone else makes. And if you are spending this kind of money on one component, you should be happy with your choice whatever it is. I just wanted to highlight that while I found the ATI "optimizations" hilarious because you just knew they were up to it too, if I can get the same quality from either I am fine with optimizing.
I just hope readers enjoy their cards and USE them, since they get old so very fast.
$.02
[Posted by: Anemone | Date: 05/27/04 12:16:55 AM]
5.
The only bone I have to pick with nVidia and why I would choose ATi over them is because of application specific optimization. If they could improve their algorithm to deliver better IQ comparible to ATi's and make it run for all games I would be more satisfied. Having the option to disable it would not be an issue if it were as good as ATi's algorithm because the difference is nothing you can notice without going into Photoshop.
[Posted by: tfranzese | Date: 05/27/04 03:02:37 AM]
6.
The test from xbit is suspicious, since nvidia is better with opengl or if you want quake3 engine based games (they are good in specific optimizations). But D3D ati is better, but nvidia is better running 3dmark (after they done the specific optimizations), but 3dmark isn't a game, so no good here. Almost forgot Unreal game engine also have the nvidia specific optimizations but you can’t disable them and it default’s to "brilinear".
See this Nvidia Anti cheat script and look at the performance numbers difference!
http://www.3dcenter.de/artikel/ati_nvidia_treiberoptimierunge /index4_e.php
[Posted by: I | Date: 05/28/04 04:40:09 AM]
[1-6]
You must log in to add comments.
Unfortunately, the old registrations do not work anymore. Please register again. We apologize for the inconvenience.
Forgot password? Registration