I was talking about for a single chip. There was a lot of interest in Cell in the HPC market, but it also died. For applications that can tolerate the latency,

I was talking about for a single chip. There was a lot of interest in Cell in the HPC market, but it also died. For applications that can tolerate the latency, offloading computation to the GPUs is such a multiplier in performance that they have to put up with manually managing the hierarchy - but even there, it's a rather coarse memory management.