William Robb wrote: > Hence my question about SLI, on the theory that I could ameliorate > that potential bottleneck [memory read/write access].
Nah, that's not what SLI is all about. The GPUs share the same frame buffer, but divvy up the work of filling it amongst themselves, when in SLI configuration. The point of SLI is that the 3D rendering computations are the bottleneck rather than the access to the frame buffer memory. In other words, it takes longer to figure out what color to set for each pixel than it takes to write that color into the frame buffer. So you break up the load and get multiple computational units working on the problem to try to "load level" between computing the colors and writing them to the frame buffer. Something could be done to improve read/write access, but it would be a different approach than what SLI does. -- Thanks, DougF (KG4LMZ) -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
