GPUs are increasingly optimised for machine learning workloads, but they aren't really designed for it on an architectural level and still leave a lot of potent

GPUs are increasingly optimised for machine learning workloads, but they aren't really designed for it on an architectural level and still leave a lot of potential performance and efficiency on the table. Current ML implementations are still quite kludgy and path-dependent, with lots of technical debt. There is huge potential for optimisation throughout the stack.