Post

AI designing its own accelerator is the loop closing in real time. RTL plus compiler plus profiler in one open repo is a serious scope. Curious how it benchmarks against a cheap commercial card for the same models.