3 ms·Wouldn't that be about as slow as CPU inference?by Havoc 3y agoWouldn't that be about as slow as CPU inference?angarg12 3y agoProbably much slower, due to all the memory thrashing and lack of pipelining optimizations.