Does nivida have plan to support direct compute shader (computation) on SLI?

Does nivida have plan to support direct compute shader (computation) on SLI?

Thanks.

I think they already do, at least in afr rendering.

But it’s only for rendering, am I right?

My question: Is Nvidia going to support directcompute shader on SLI? For example, I have two grahic cards connected with SLI. Then I run a directcompute shader. Can the compuatation be dividely evenly so that each card only takes half of the computation?

Thank you.

No, it is for compute shaders as well, when the code also calls the present method with sli configured in afr mode (which is the default SLI mode).

No, but that isn’t how modern SLI works anyway. Split frame rendering is basically dead, and the modern Direct3d API is really designed only for alternate frame rendering into the render buffer.

Yeah, we support DirectCompute with SLI in AFR mode. You have to be careful about dependencies between frames (same with graphics).