3 ms·
Yes. Take a look at, say, CUTLASS: you'll see that they use PTX instructions because there are no intrinsics, much less automatic compiler lowering, for the acc
by saagarjha 2y ago
Yes. Take a look at, say, CUTLASS: you'll see that they use PTX instructions because there are no intrinsics, much less automatic compiler lowering, for the accelerators they target.
- HarHarVeryFunny 2y agoYes, but that's an NVIDIA project, so would be expected to be hand optimized, same as their cuDNN kernels. I'm more curious about what types of model people in research or industry are developing, where NVIDIA support such as this is not enough, and they are developing their own PTX kernels.