4 ms·
I learned about this recently when trying to optimize ML test architecture running on Azure. It turns out having access to Ice Lake chips would allow optimizati
by hallway_monitor 5y ago
I learned about this recently when trying to optimize ML test architecture running on Azure. It turns out having access to Ice Lake chips would allow optimizations that should decrease compute time and therefore cost by 20-30%.
- bee_rider 5y agoSome AVX-512 stuff I guess? AVX-512 had a rough rollout, but it seems like it is finally turning into something nice.