Comment Re: Please explain (Score 1) 6
The point is that it *wasn't* a solved problem for machine learning / deep learning workloads in GPUs, which are very different from conventional workloads on CPUs.
Of course there's no reason for it, it's just our policy.
