I really beg to differ. Obviously a 512/1 Gigabyte 9800GTX is a more perfect comparison. But the tests are hardly invalid. Especially when the problem is obvious. Medion you should know me well enough by now to know that I am very aware of what a memory bottleneck is and what an SLI Scaling problem is. Yes there can be other factors. But in this case its very clear what the bottleneck is. And its the same thing with all those tests all failed 16xQ tests. The only common factor between the 8800GT, 9800GTX, and 9800GX2 and those failed tests was the 512 megs of memory. Your not suggesting that this is some sort of coincidence.
I think its pretty clear where the bottleneck/problem was. In my old 9800GTX 3 way SLI/9800GX2 Quad Comparison to 3 way 8800GTX I had the same problem with Crysis. I've done extensive testing with this.
Originally Posted by myself
Editors Note: With soft shadows set to high the 9800GX2 and 9800GTX configurations all fall down to 5 FPS range and the 8800GTX Tri SLI Configuration is able to hold a 22 FPS advantage. Even with that advantage the 8800GTX 3 way SLI configuration has unbearable mouse lag due to the low performance of the single GPU. However it should be noted that this is one spot where the 8800GTX's extra memory gives it an advantage.
Now I performed these tests well before I even heard of an "HD4000" card. Now either I'm lying or making this up. Or I do know what I am talking about and these problems do happen on 512 meg cards. I have consistently shown that 512 meg cards have problems and the problem has grown with newer and more modern titles. Yet you are quick to dismiss it with /yawn different configuration type posts. I've gone through a great deal of trouble here to explain my position and the experiences/tests I have done.