hartz
1 Dec 2005, 9:44am
If the work of rendering a frame on the screen can be split over 2 GPUs, why can't it be split over 3 or more, eg each card doing a 3rd of the image using some kind of sizzor mode? Why not 10 GPUs each do a tenth of the work?
And I don't want to start any conspiricy theory discussions, but why all the [chipset] lock-in? SLI only works on an nForce4 board (bar the K8T900 reference board tested here: http://techreport.com/reviews/2005q4/via-k8t900/index.x?pg=1) , and Crossfire only on an ATI chipset board ... S3 is doing the right thing by "just using the PCI-E bus for MultiChrome", irrespective of wat motherboard you use. The bandwidth is there, the bus allows communications, the software takes care of the workload distribution. What else do you need?
And I don't want to start any conspiricy theory discussions, but why all the [chipset] lock-in? SLI only works on an nForce4 board (bar the K8T900 reference board tested here: http://techreport.com/reviews/2005q4/via-k8t900/index.x?pg=1) , and Crossfire only on an ATI chipset board ... S3 is doing the right thing by "just using the PCI-E bus for MultiChrome", irrespective of wat motherboard you use. The bandwidth is there, the bus allows communications, the software takes care of the workload distribution. What else do you need?