getWarpsPerSmBw Class — pytorch Architecture
Architecture documentation for the getWarpsPerSmBw class in kernel_backward.h from the pytorch codebase.
Entity Profile
Source Code
aten/src/ATen/native/transformers/cuda/mem_eff_attention/kernel_backward.h lines 193–200
template <typename scalar_t, typename Arch>
constexpr int getWarpsPerSmBw() {
bool is_half = !cutlass::platform::is_same<scalar_t, float>::value;
if (Arch::kMinComputeCapability >= 80) {
return is_half ? 12 : 8;
}
return 8;
}
Source
Analyze Your Own Codebase
Get architecture documentation, dependency graphs, and domain analysis for your codebase in minutes.
Try Supermodel Free