+1
The issue with Bit-Tech/Custom PC is that they end up with the absurd conclusion that:
- GTX550 Ti is worth buying over the 6850
- 2GB VRAM on a 6950 is not only pointless, but also the GTX560 Ti is faster and cheaper
- 6970 is slower than the GTX570 and more expensive
- GTX590 is faster than the 6990
Basically they said "buy nVidia, because everything else is rubbish."
They also only test four games - BC2, BlOps, Dirt 2 and Arma II, which, frankly, two of those are Nvidia favourites, one is irrelevant (BlOps - almost any card will produce good FPS here) and Arma II is ambivalent.
Take the GTX590 vs 6990 for Arma II at 2560x1600:
Card____Min__Average
GTX590: 31____38
6990:___26____49
And the conclusion is that the GTX590 is a faster card. Without being able to say how often it drops that low (and judging by the average, not often), the minimum result is irrelevant. If they could provide a max FPS too, it would help decipher that element to a degree - if the average and max FPS are very close (say a max of 51 fps for the 6990), then it's likely the min fps was just a random drop and bares very little relevance, whereas is the average and max FPS were miles apart (say a max of 91 for the 6990), the game would likely feel choppy. Even then, without time information, these conclusions could be inaccurate.
I think a line graph of the FPS over time would be helpful, but comparing many cards in such a way is difficult, as the graphs end up very confusing and busy very quickly.
EDIT: It looks like BT have corrected the GTX590 result since their 'graphics card megatest' in CPC, as now the averages are half-decent. But to come to the conclusion that the GTX590 is faster, even with driver problems (I guess that's what accounts for the difference) is ridiculous.
