| 3DNews Vendor Reference English Resource - All you need to know about your products! |
||||||
![]() |
||||||
|
|
||||||
Gigabyte GeForce 6800 Ultra vs ASUS AX800XT: total DeathmatchAuthor: Anna TimofeevaDate: 26/09/2004
Introduction
NV40 / R420. This combination of digits/letters can be called a live legend. Before the official release there were so many talks, hopes, sometimes incredible assumptions around. All were astonished when these products were finally presented to the public, and the results of first tests announced. And the reviewers... These "tears invisible to the world" when you can't believe your eyes on seeing the FPS output. Is it reality or an illusion, "Matrix" hallucinations?
All this is behind. First delights are over, those who fainted awakened, and the bodies of those who failed to come to the senses were brought out. Initially raw drivers seemed to be brushed up. All started to think, hardly finding their ways in new video card lines and lots of specifications that fell upon the readers' heads from all hardware editions. Now it's all behind. What is currently going on used to be dubbed as "interlunation". Cards have been released, all are about to guess what each looks like, some even have got them on hands (such semi-mythical persons enjoy an immense popularity on forums of the respective topics), but at the same time the high price (we are now talking only about ultra high-end cards) and the scanty quantities of the boards in the retail are serious obstacles in the way to the cherished box. But those who are going to buy such a video card for their home computers (simply, the vast majority) are excited and pleased to follow the fight for leadership "at the Olympus summit". This "third person view" can be compared to how fans are watching the games of their favorite teams in the Champions League over TV. Yes, they do see (the test results) with their own eyes, therefore they can produce own impression, which sometimes differs from that of the arbiter. Also, during the whole match we always hear the commentator's voice (the reviewer's comments) who distinctly and as impartially as possible assesses the situation that has shaped up (because football is a very dynamic game, so are computer games) thus assisting the audience (the reader) in finding their ways, analyzes the situation on the field (on the video cards market) that's currently going on in the market creating the effect of own presence, at least, on the stadium stand. We have now presented to you a description of the ideal commentator. It's up to you to decide on the analogies in this regard. But it's hardly arguable that this comparison fully expressed the essence of what is going to be today on these pages. Two products of the ultra high-end products, a major advance in science and technology, will be competing today for the title of absolute champion with Catalyst 4.8 and ForceWare 61.77 drivers that has gained the latest WHQL certification. The Nightmare tier. No kidding. ![]() We won't be repeating for the hundredth time the technical specifications describing the specifications for each solution in detail. To refresh those detail in memory, just look at the following table:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| CPU | P4 3.2Mhz 800FSB (Northwood D1) |
| Mb | Epox 4PDA2+ (i865PE) |
| Memory | PC2700 2x256Mb 400Mhz in the dual-channel mode Latency timings - 2.5-8-4-4 |
| VGA | Gigabyte GeForce 6800 ASUS AX800XT |
| OS | WinXP + SP1 |
| Drivers | ForceWare 61.77 Catalyst 4.8 |
Gaming benchmarks:

We are starting a Deathmatch in the so familiar and repeatedly run 3DMark 2003.
These are the first maps. Even with the complexity of Nightmare, they are quite easy to walk through, and it doesn't make much sense making global conclusions about the rivals' ranking and the alignment of forces.
However, the preliminary totals are such that 6800Ultra won the first round.
Õ800ÕÒ took the second, aka the last, place. But the gap is not that wide to talk about a complete knock-out. Running ahead, I am telling you that X800XT will win back, several times. Now let's look at the detailed alignment of forces over the tests:

Benchmark No.1. Used are vertex shaders of ver. 1.1.


Pixel shaders of ver. 1.1 and 1.4, Dynamic Shadows and Depth Of Field effects.

But in the Nature benchmark fully made up of pixel and vertex shaders of ver 1.1, 1.4, and 2.0, Õ800ÕÒ delivers a heavy blow.





In 3DMark2001 SE, the situation changes to the reverse;) 1:1.




Under 4õÀÀ and 8õAF, the Õ800ÕÒ looks more attractive. However, this is merely synthetics. In what follows, there go more complicated tiers. Gaming benchmarks. Only there it will be fully clear who will get more frags in the end.

This is a semi-gaming semi-synthetic benchmark where the major accent is made on the pixel and vertex shaders. 6800Ultra wins back completely.
At HSR test, both cards exceeded 200fps, but as is known, VillageMark does not understand results higher than this bar...


A demo version of the game "X2 - The Threat". This benchmark is quite hard for video cards, at which both rivals make quite a good showing.

A complete parity. The results differ within measurement errors.



It seems like a bottleneck was found in 6800Ultra. It doesn't like high resolutions in the demanding quality modes. It was absolutely useless testing these cards without AA and AF and, needless to say, with the Low Quality settings, because both these solutions have "overgrown" most benchmarks in which their performance is tested. That is seen even on the above graphs.

This benchmark that uses mainly pixel and vertex shaders of ver. 2.0 at which NVIDIA is normally a winner has not let down either - it's no wonder though.

It's up to you decide whether to believe the results of this benchmark or not. We simply bring in the figures. We are looking forward to the time when we can test the card not with this shabby demo that sometimes shows quite dubious results, but with the final version of the game. Actually, it is not improbable that we'll get the same picture which we now observe in DOOM3 (but in the reverse), of which we already wrote.
But anyway it is good that GeForce 6800 Ultra was tested with this demo. In the previous versions of ForceWare drivers, NVIDIA cards of the sixth series for this application were locked, like about other five tests. From this list, only TRAD is left:




This is already a more serious battlefield for such advanced contenders. It is a game of our time which can be regarded as a worship game. Incredibly popular among the gamers, which is quite justified considering its gameplay and the nice scenery it gives. But you can fully enjoy the scenery only with video cards of the respective class, at the maximum quality settings. All these conditions were strictly met, and judging by the results we see that the rivals are worth of one another. Remarkably, the use of 4õAF does not affect the graph result.

That is a test that came from the DX7 epoch. We used it only as a tribute to traditions. Once again, the heavy lag at 8xAF against the background of overall prevalence of 6800Ultra appears to be an alarming architectural signal.

At UT2004, we see a picture similar to that in UT2003. We are ignoring the graphs produced under the Low and High settings without aggressive quality settings because they are absolutely uninteresting in terms of performance comparisons of two super-top solutions - there are merely horizontal lines for them.

One month ago, exigent readers waited for the Far Cry results as a dessert, and now results for DOOM3 are in the same demand. All is predictable enough - the NVIDIA card offers comfortable gameplay in Hi Quality up to 1280x1024 (70 fps), while that of ATI's - up to 1024x768.
Since half the office enjoy playing DOOM3 at home with quite standard computers on R9600XT, then it's an immense happiness when there is a chance to get at a spare VGA monitor, 1GB RAM and a 3.2GHz processor! You take off something like 800XT or 6800Ultra from the shelf and enjoy the gameplay in DOOM3 with a 21" monitor getting the cherished 70 fps without any delays at all :-).
We have already prepared some curious figures for another review, and can't help sharing them with you. Unfortunately, our "guinea pigs" are not reflected in this graph. I would call it exotics of the PCI-Ex class, so it is of only academic interest just to demonstrate the huge performance gap for one and the same card run with a "prehistoric" driver (when there was no DOOM3) and the most recent WHQL-certified driver:

This part is the most puzzling. Those who expected to get the findings like "X finally knocked out Y" or "X lost to Y without a score" are in for a disappointment - the cards are are about equal and demonstrate fantastic performance with their already mature drivers for which there is no reason to be ashamed of. I believe, those who simply wanted to have a nice time watching the fierce struggle in the ultra high-end sector have got a real pleasure reading this material.
It goes without saying, we did not expect anything radical or shocking from the test results. Proceeding from the situation on the market that we observed before the release of NV40/R420, we could foretell that NVIDIA simply couldn't help producing something that immediately makes you forget about the history with NV30, but ATI in turn can't afford to drop speed and rest on the laurels.
In fact, that has come true as we expected. There is absolutely no "total domination" in the top sector. Some sort of parity has been established, which is to the benefit for the consumer - the war should switch to the price niche. Meanwhile, we shouldn't forget that these "masterpieces" require respective frames - simply speaking, hardly can you expect a fantastic performance boost from Õ800ÕÒ or 6800Ultra on a Celeron-powered PC. The system has to be well-balanced, which also applies to a comfortable gameplay.
Now regarding the games. The very powerful and sharp upward leap on the market of video cards has resulted in the most games turned out to be "piece of cake" tasks for new video cards. Of course, nature abhors a vacuum. Take for instance the Ultra Quality mode in DOOM3 which a priori forces even the most confirmed gamer to upgrade the card in some time even though the card is a super-top :). But that relates more to the "psychological pressure" the marketing strategy is notorious for in varied fields. It's no use setting hopes for that, because it was, is and will be that way.
As regards the price bar, there is some sort of parity at that. The recommended price for 6800Ultra and Õ800ÕÒ to date is approximately equal - 500$. Note also that the price is recommended, and it may radically differ from that shown in the price list of a neighboring shop. In this case it is useless and illogical to set hopes for NVIDIÀ or ATI. Firstly, for ultrahigh-end solutions the production is more complicated technically than in the middle-end sector and thus costs an order of magnitude higher.
Secondly, the number of such cards in the retail is scanty. To date, this topic is urgent as never before, and according to all the market laws this fact provokes some pricing mess. For now, both these cards are in great shortage, but we held the first samples in the hands as early as in April!!! (the review was released in May after the NDA period expired). There hasn't been such a delayed announcement in history yet.
But let's not talk about the sad. Today we have "touched the legend", which brought us an immense pleasure, because as every fan of their business we are delighted to deal with hardware, especially so perfect, and explore its features in detail. Before us, we face further stages of progress, new video cards, new tests, but in any case the period of running "6800Ultra VS X800XT" Deathmatch at the very summit of video accelerators is already a thing of the past history.
| |||||