3 ms·
Volta V100 already has "tensor cores" which are basically little matrix multiplication ASICs.
by deepnotderp 9y ago
Volta V100 already has "tensor cores" which are basically little matrix multiplication ASICs.
- Nokinside 9y agoThat's what I said. The microarchctiecture has many unnecessary things and it's not optimized as a whole for deep learning.
- deepnotderp 9y agoI believe it was either the last MICRO* or the one before that when Dally addressed this point. The specialized hardware for graphics ends up comprising such a small portion of the overall chip that it wasn't worth it to remove it. The "GPUs were made for graphics thus aren't good for DL" argument really doesn't hold a lot of water IMO. * It might've been a difference conference now that I think of it.