MatID/Z-depth render seperate frames on individual gpus?

[GR1F1TH]

I was sitting here testing out the new Kernel for rendering Material ID's and couldn't help but think that these "quick passes" would run even quicker if it were possible to load a frame per gpu in multi-gpu settings. I my case I have 3x gtx465 and get a pretty clean pass at 256 samples in 1-2seconds. The hold up (although extremely quick) is the load time. If three frames were sent out at a time (one to each gpu) this would seem to save a large chunk of time for things like rendering a material ID or a z-depth pass. Incorporating this might be too much effort for too small of a gain for too small a user group but I'm one of those people and thought I'd throw this observation out there.