GeForce FX catches up to Radeon performance

Dom

The latest update of the GeForce FX adds a significant performance boost, and appears to approach the Radeon 9800, according to ZDNet Labs - though Nvidia has been accused of fiddling the tests
Nvidia's latest high-end graphics processor appears to have caught up to ATI's latest Radeon chip in performance, even allowing for alleged dirty tricks in the way Nvidia's drivers handle benchmarks, according to new tests from ZDNet Labs.

(See the full review: Nvidia GeForce FX 5900 Ultra: the first benchmarks.)

Nvidia and ATI are battling for the right to call their technology the fastest, a right which carries prestige with high-end gamers, even if it does not translate directly into sales in the more lucrative mass market.

The new set of benchmarks pitted the new Nvidia GeForce FX 5900 Ultra against ATI's latest, the Radeon 9800 Pro 256MB. Nvidia holds a substantial market share lead over ATI, but ATI has until now held a performance lead in the current generation of GPUs (graphics processing units).

ZDNet's tests found that Nvidia's improvements to its new chip have given it an edge over the Radeon in some areas, such as 3DMark03 benchmark tests. In other tests, such as parts of the Unreal Tournament 2003 test and the Tom Clancy's Splinter Cell test, results were more mixed or the lead went to the Radeon.

The results are a big improvement over Nvidia's first iteration of the GeForce FX, the 5800 Ultra, which was delayed and then showed disappointing performance.

However, some uncertainty hangs over the benchmark tests, since Nvidia has been accused of tweaking the way its drivers handle benchmarks in order to give an artificially high score. Even so, ZDNet's testers concluded that Nvidia has at least caught up with ATI in terms of sheer 3D performance.

________________
Source: ZDnet

means96

I just saw this at Guru 3D

Okay, so this is the last you will hear of us regarding the topic as it is getting into this weird mud-fight, now if there would be any girls in them I'd welcome it

For the record. Based upon the opinions you guys broadcasted and through what I see in the poll the entire staff decided unanimous it would be best not to use 3DMark 03 software in our benchmark suite for the time being.

For the bigger part of it you should not blame FutureMark for this though, but blame the parties that started cheating. These are both nVIDIA and ATI and I don't care wether its a 2% or a 25% difference, cheating is cheating. I actually applaud nVIDIA for the way they did it, if you do it then have the b@lls to do it well.

The results now and in the past few months have been rendered useless. I expect that future drivers again will get optimized and we will simply not be any part of that as you are buying products based upon our reviews.

Now then, hardavenue got the skinny on FutureMark's take on nVIDIA's response.

Any suggestion that Futuremark would intentionally penalize or favor any specific hardware in our products is absurd and NOT TRUE. And please note that we are not attacking here, we are just defending our product.

We are in the business of making objective benchmarks and we intend to keep it that way. We respect deeply our BETA members, which include the biggest in the industry, and they all have participated in the development of 3DMark03. nVidia themselves was an active member of the BETA program until December 2002.

NVIDIA's claim that "since they are not a beta partner, they do not get a chance to write shaders like they would with real applications developer" is irrelevant and to us it seems like an attempt to shift discussion to a different topic. The topic here is that the drivers special cased 3DMark03 and resulted in an incorrect score which made false representation of their products' performance in 3DMark03.

3DMark03 was developed strictly according to DirectX9 standard in very close cooperation with Microsoft and other BETA members. If hardware performs well 3DMark03, it performs well in all applications that use DirectX 9. Note that since 3DMark is designed to be an objective evaluation tool, it does _not_ include manufacturer-specific optimizations. This is I just saw this at Guru 3D

why it is exceptionally well suitable for objective performance measurement.

Since we all now see how tempting it is to try to cheat in a benchmark, we have started to work with our partners in order to develop new structures and processes to weed out cheating and unfair play in all benchmarks. This is a business opportunity for us, since due to our role and position we are better suited to act in this role than e.g. game benchmark providers. We welcome all interested parties to work with us in this important initiative.

nVidia is an extremely capable company with great products. We welcome them to continue the competition in the hardware development with fair means.

In reply to your more specific question, 3DMark03 is a forward looking DirectX 9 benchmark. It stresses the hardware with workloads that will be typical of DirectX 9 games. Thus, hardware's performance in Unreal Tournament 2003 does not necessarily bear relation to that how that hardware will perform in DirectX 9 workloads, which make an extensive use of pixel and vertex shading. Doom 3 is not a published product yet but we have seen few reviews. Based on those reviews, it seems that in Doom 3 there are performance differences depending on which codepath is used for different hardware. Not much more can be said at this time, as the application is not ready and not available for public testing (although we very much would like to see it
As stated .. for the time being 3DMark 03 will no longer be part of our benchmark suite. We'll monitor things closely though, and if it's safe to use the software again we'll definitely use it.

nForcer

If only more people saw it this way.

I wish they would stop going for the fastest and shoot for the best quality. If I remember right, sometime when the 5800 was due to be released, an nVidia Rep said, "Sure 200 fps in QuakeIII is nice but whats the point?"..."Why not improve the quality of the game instead of trying to achieve a frame rate that the human eye can't notice?" (don't quote me on the exact words... can't find the url in my bookmarks)

I'm a diehard nVidia fan (until they give me a reason not to be) but I wish BOTH ATI and nVidia would wise up and call a truce. That or try for different bragging rights "Best Quality GPU" instead of "Fastest" because the current 'competition' is reaching its limits - and reputations are being tarnished. Not to mention a good number of gamers are just completely fed up with the whole mess. (which I'm sure we can all agree)