| Review index: |
|
|
Benchmarks in a nutshell
We calculated the average of all benchmark results and gave the same weight to all games:
On averages performances vary by as much as 5 times in 1280 x 1024, but only as 3 times in 800 x 600 as in this mode the most efficient graphic cards are restricted by the CPU. For middle range products, the 6600 GT, 6600, 6600 LE versus the X700 Pro, X700 SE and X600 Pro NVIDIA wins. The performance gap isn’t huge, but it is there with as much as a 15% advantage for NVIDIA.
It is also possible to see the X600 Pro as a direct opponent of the PCX 5750. It was the case a year ago for the 9600 Pro and FX 5700. At that time NVIDIA, was working hard on drivers to reach an acceptable performance level. It’s no longer the case today and as we can see from results: the PCX 5750 is completely left behind by the X600 Pro.
For entry level products, it’s more complicated as products aren’t exactly on the same level except for the TurboCache 32 MB and HyperMemory 32 MB, which provide equivalent performances with a slight advantage to NVIDIA. These two cards are very close to their 128 MB 128 bit equivalents, proving the interest of TurboCache and HyperMemory technologies. TurboCache 16 MB is one step behind as 16 MB is really insufficient. The TurboCache 64 MB is just slightly behind the 32MB version, showing the importance of bandwidth over the quantity of memory.
The 6200 and X300 provide similar performances with a slight advantage to ATI this time. The Radeon X550 is one step above with performances close to the X600 Pro.
The i915G and i945G performance level is lower than that of all ATI and NVIDIA graphic cards except for the 6200 TurboCache 16 MB, which is surpassed by the i945G. This doesn’t bring any value to 3D performances.
Practical tests in a nutshell
To calculate the average graphic quality reached for 19 games with comfortable playability, we gave 1 point per level of quality. For example, if there are 5 resolutions between the best and the worse graphic card, the cards obtained from 1 to 5 points according to their capacity (and 0 if play wasn’t possible). We then leveled all results for each game to have the same weight and made an average:
First of all, these results confirm the standard benches for all middle range graphic cards and partly so for entry level products. Results are similar between ATI and NVIDIA products.
For bottom entry level products there are a couple of differences. The X300 SE and TurboCache 16 MB provide similar results, a step below other entry level graphic cards. It wasn’t possible for 4 out of 19 games. Here, Intel integrated chipsets are clearly behind, whereas they were closer than entry level graphic cards in the benches. Why such a difference? Because it isn’t possible to play several games, and this resulted in several no scores. Of the 19 games 8 were unplayable, half of these because of bugs.
