Foxconn 8800GTS - even the "younger sister" of GeForce 8800GTX beats yesterday's flagships
Overclocking
This time, we are not bringing in the results of tests for Foxconn 8800GTS when overclocked. It's not because we did not try to overclock the video card. We did. But look at the below screenshot - these frequencies match those for 8800GTX and not 8800GTS!
While this is quite possible for the frequency of the video chip, then the nominal overclock of video memory from 1600 to 1800 MHz DDR is very unlikely. On the other hand, Foxconn has always followed NVIDIA's official recommendations. Although PowerStrip correctly detected the name of the video card, it seems that the utility is not very well familiar wit the video card. Nevertheless, we can move the slider bars and Foxconn 8800GTS is responsive to that - the results go up vividly. Nevertheless, we are not bringing in the results produced through overclocking because if the frequencies of Foxconn 8800GTS are nominal 500/1600 (which is most probable), then it is impossible to say at which frequency we produced the results for overclocking. As you will see in what follows, even at the nominal frequencies Foxconn 8800GTS is able bowling out the competitors.
Regarding the competitors. As contenders to Foxconn 8800GTS, we took GeForce 7950GX2 and Radeon X1900XTX. The former was quite recently regarded as the fastest "single" video card in NVIDIA's product line, whereas Radeon X1950XTX on the moment of tests was unavailable, unfortunately. So we had to make some replacement in the team, and its place was taken by Radeon X1900XTX. The performance difference of these video cards amounts to 3-7% depending on the application, so it is possible to perform comparisons without loss of generality. Why are we comparing 8800GTS just versus these video cards? Because, as we already stated, 8800GTS albeit "a young sister", its retail prices are still biting and even are higher than that for the dual-chip 7950GX2. That is why we'll try to answer the vital question – does it really make sense to urgently migrate to NVIDIA's new architecture or to wait a bit expecting the release of DirectX10 games?
Test setup
|
Bus
|
PCI-Express
|
|
CPU
|
|
|
MB
|
|
|
Memory
|
|
|
OS
|
WinXP + SP2 + DirectX 9.0c
|
|
PSU
|
Thermaltake ToughPower 750 W
|
To test Foxconn 8800GTS, we used ForceWare 97.02, and for GeForce 7950GX2 – ForceWare 93.71 WHQL, for Radeon X1900XTX – Catalyst 6.10.
Benchmarking
We decided to slightly change the look and feel of the diagrams. Now the results don't go width-wise, but upwards. Besides, on a single diagram we bring in results for three resolutions at a time in order to estimate the rate of change in results as load to the video card changes. The results for Foxconn 8800GTS are highlighted in bright green, GeForce 7950GX2 - in dark-green, and for Radeon X1900XTX - in red. As regards the testing modes, in each of the games we chose the maximum quality settings, wherever possible. The make the load upon the video cards harder, we enabled the AF16x and AA4x modes. This time, we did not enable the FSAA settings available to GeForce 8800GTS, because that is a separate topic which will be some time afterwards.
Well, let's go ahead. As usual, let's start with the synthetics.
The very first test shows that 8800GTS is on par with the dual-chip 7950GX2 whose GPU runs at the same 500 MHz. Moreover, as the load in terms of number of complexity of shaders goes up, this adds more to the advantage of 8800GTS. The Radeon is almost one and a half times behind 8800GTS. Mind you, some "younger sister"!
The next series of tests in shooters was produced with the SmartFPS utility. This utility allows to measure the performance in six popular games in the automatic mode. It is available for download from out file archive.
The graphic engine of Call of Duty 2 is known for its clumsiness. At this test, 8800GTS failed to overtake GeForce7950GX2, and even Radeon X1900XTX does not lag behind that much. Does that mean it makes no sense in upgrading the yesterday's top-ends to 8800GTS? Let's not be hasty with conclusions.
But at Battlefield 2 the novelty is much more comfortable and overtakes both the predecessors. In fact, there are some points to complain about this demo. It is seen that even in the 4AA/16AF mode at low resolutions the results are bounded by the CPU performance. We'll sort it out what was the case - is that a relatively old engine or the scene used in the test.
At Serious Sam 2, the Foxconn 8800GTS takes a sure lead…
… But at Quake 4 it holds firm positions between Radeon X1900XTX and GeForce 7950GX2.
As regards the last two games included into the SmartFPS pack, then at both Prey and The Chronicles of Riddick: Escape From Butcher Bay we see a very similar pattern of results. Radeon X1900XTX is well behind the "greens" and they run abreast, except that GeForce 7950GX2 appears to be a little bit ahead. For instance, compare it versus the results for DOOM3.
At DOOM3 the difference between GeForce 8800GTS and the leading 7950GX2 is somehow greater.
At F.E.A.R., the overall pattern preserves. Except that Radeon X1900XTX is coming closer to the leaders.
We can see an interesting trend – at relatively "old" games or games based on old graphic engines the GeForce 7950GX2 takes a lead, while 8800GTS is running a bit behind. At much later games abundant with complex shaders the superiority is with GeForce 8800GTS.
That can be seen in the spectacular Company of Heroes. The gap between GeForce 8800GTS and 7950GX2 is not great but distinct enough and becomes wider as the resolution goes up. But Radeon X1900XTX does run well behind.
The next games were included into the set of tests following the numerous requests from our readers. We thank all who took part in the discussion on the forum. The games that follow do not have integrated performance meters, so we used the FRAPS utility. Don't be hasty at sneering skeptically about the trustworthiness of results which we produced with the FRAPS utility. Some time before the announcement of G80, we started developing a methodology for testing performance of video cards at games which don't have an integrated benchmark, and succeeded in that. It happened that currently the text of the methodology requires some "brushing-up", that is why is has not yet been published on the web site. We hope our readers will forgive us for this shortcoming which we'll fix in the nearest future. But for now let's see what we got in the end.
The first number of our show is the very popular arcade game - Need for Speed Most Wanted. Although the game was released a year ago, it still gives a substantial load upon the video system and is quite demanding for resources, especially at the maximum graphics quality settings.
We immediately see a sensation. The 8800GTS leaves no chances to its rivals! Despite the maximum quality settings, the decrease of results in 8800GTS with the rise of resolution is not that intensive as observed in the recent flagships. Is that a triumph of the new unified architecture in action? Now let's look at the results for the next representative of the Need for Speed series.
Again a surprise. At Need for Speed Carbon, we see an evident failure of GeForce 7950GX2. Believe it, it's not the methodology of measurements that matters. Having come across such a strange result, we verified it several times and every time we saw the same pattern. Most likely, it is the drivers to blame for it. It is well known that NVIDIA pays much of attention to the driver optimization for each application. This time, it seems the game is "hot off the press" (have not yet been optimized), or perhaps it is not easy to get distributed through parallel streaming between two GPUs. Somehow or other, there is something in there. If the guess regarding the lack of optimization for this specific game is true, then it is easy to explain why GeForce 8800GTS no longer thrills us with the huge gap as it was in NFS Most Wanted. Quite probably, with time the results for GeForce 8800GTS at NFS Most Wanted will go up substantially. But Radeon X1900XTX demonstrates almost the same results at NFS Carbon as they were in NFS Most Wanted.
The game Titan Quest is still among the fresh novelties and is abundant with complex graphics. For a game that was released under NVIDIA's aegis «The Way it's Meant to be Played», it is quite expectable that video cards by this company will show higher results, but we did not expect they would be SO high… Just look - GeForce 8800GTS demonstrates an almost twofold superiority over Radeon X1900XTX while leaving the dual-chip 7950GX2 well behind.
Finally, the Oblivion game. At this test, we did not enable the HDR since 7950GX2 does not support this simultaneously with the FSAA activated. The remaining graphics quality settings were set as per the Ultra profile. So, what can we see now? GeForce 8800GTS runs on par with 7950GX2 and take an essential lead over the Radeon. And even the resolution 1600õ1200 is quite comfortable to play for 8800GTS.
Final Words
Let's recap it all up. In old games,as we stated, 8800GTS demonstrates quite decent results but anyway somehow lower than the previous flagship GeForce 7950GX2. In new games abundant with complex shaders, the unified architecture G80, albeit in some concise version as compared to 8800GTX, demonstrates a convincing superiority and reveals its capabilities. Does it make sense to upgrade and replace 7950GX2 with 8800GTS right now? The pros of 8800GTS - the higher performance, support for DirectX10 and thus future games, support for higher-quality anti-aliasing modes concurrently with the HDR. The cons: the retail price of the novelty may be substantially over than that of 7950GX2, especially at the time right after announcement. It is up to you to decide which option suits you. In general, the product has proved really successful. We recommend!
Discuss the article in a forum.
 |
Top Stories: |
 |
 |
|
MoBo:
| |
|
|
VGA Card:
|
|
|
|
CPU & Memory:
|
|
