Hacker News new | past | comments | ask | show | jobs | submit
aesthesia | on: Making deep learning go brrrr from first principles (2022)
That's also a CPU that came out four years later than the A100. The contemporaneous B200 is not optimized for FP32 and does 74.45 TFLOP/s. For FP16 it's at ~2 PFLOP/s.
p1esk | parent
The point is that modern CPUs are not as slow as most DL people think. Roughly 10x slower but with a lot more memory.