In Photogrammetry, at least, some programs have codepaths that run concurrently: part heavily parallized for CPU, and part heavily parallized for GPU. I built a

In Photogrammetry, at least, some programs have codepaths that run concurrently: part heavily parallized for CPU, and part heavily parallized for GPU. I built a monster system with dual 1080s when they first came out, and my i7 became the bottleneck. There's algorithms that haven't yet been optimized for GPU parallelization. And there's datasets that require memory access patterns that aren't amenable to GPU. That said, now I'm curious what a 128-core system would do to memory access. Better hope the algorithms are L2/L3 coherent?