dual-processor video cards in benchmarks
Hi all -
just looking up specs on my new video card - haven't submitted myself yet - but I see something that seems VERY strange. Why are the brand-new dual-CPU ATI & nVidia cards ranked so low?
The Radeon 6990 (my new card!) and the GeForce 590 should both be at that top of the video benchmarks by far, should they not? Seeing as both are essentially dual-CPUs of a slightly-lower single-CPU card. So why is the 6990 rated barely above other single-chip ATI products, and the 590 is even BELOW other nVidia products by a large margin! According to the existing scores, the 6990 is barely 3% (!) faster than a 6970 which costs 1/3 the price, and a 590 is almost 20% slower than a 580..
Is this a benchmark failure to exploit the capabilities of dual-chip cards, or a driver failure to activate both chips during testing, or is the user's system to blame somehow (not pairing a top video card with an adequate CPU or board?)
See these old posts.
From our point of view it is mostly a device driver issue. But as mentioned in the posts above, we'll have another look at SLI / Crossfire in the next major release of PT.
Yes, it has been a long running issue. We tried to contact them (ATI and nVidia) a couple of times. They made some positive noises, but nothing happens. We also tried nVidia SDK (to force particular SLI modes) but it didn't improve performance. In fact it hurt performance, or corrupted the display in some cases when we forced SLI.
The issue is the same for dual cards or dual chips on one card.