CUTLASS 3.5.1 (#1623)

* CUTLASS 3.5.1

* updates, optimizations, fixes
This commit is contained in:
Vijay Thakkar
2024-07-29 08:46:24 -04:00
committed by GitHub
parent 56b46e2d13
commit be60a0b272
312 changed files with 19793 additions and 6775 deletions
@@ -231,7 +231,7 @@ struct B2bFusedGroupedGemmRun
host_tensor_ref_D1.at(i).sync_device();
ref_A0.at(i) = (host_tensor_A0.at(i).device_ref());
ref_B0.at(i) = (host_tensor_B0.at(i).device_ref());;
ref_B0.at(i) = (host_tensor_B0.at(i).device_ref());
ref_C0.at(i) = (host_tensor_C0.at(i).device_ref());
if (alpha0 == ElementCompute(0)) //per-channel scale
ref_Scale0.at(i) = (host_tensor_Scale0.at(i).device_ref());
@@ -340,7 +340,7 @@ struct B2bFusedGroupedGemmRun
std::cout << "Fusion time " << gemmTime / (float)runs << " ms\n";
for (int i = 0; i < problem_count; ++i) {
host_tensor_D1.at(i).sync_host();;
host_tensor_D1.at(i).sync_host();
//
// Verify