I think the question depends on the assumption that GPU's share workloads, which they do not.
Mining software works similar to the way that parallel computing clusters do whereby the software uses a small amount of the host's CPU resources to assign jobs to each GPU, each of which are handled by one GPU.
Thus, each CPU works one job at a time and the work of one job is not shared between the GPU's.
This is why it doesn't matter if your GPU's are in Crossfire or SLI configuration (or not). The software addresses the GPU's directly.
You could have as many different GPU's as your motherboard can support all running under one instance of, say, Diablominer.
Mining software works similar to the way that parallel computing clusters do whereby the software uses a small amount of the host's CPU resources to assign jobs to each GPU, each of which are handled by one GPU.
Thus, each CPU works one job at a time and the work of one job is not shared between the GPU's.
This is why it doesn't matter if your GPU's are in Crossfire or SLI configuration (or not). The software addresses the GPU's directly.
You could have as many different GPU's as your motherboard can support all running under one instance of, say, Diablominer.
So say I have a Sapphire Radeon HD 7950 and I want to buy a PowerColor Radeon HD 7970. Would that work?
