leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 24 Packages Projects Releases Wiki Activity

releaase 2.11 (#703)

Browse Source
This commit is contained in:
Aditya Atluri
2022-11-19 09:02:15 -05:00
committed by GitHub
parent 3c90f6aea6
commit c975e2ccbb
329 changed files with 47332 additions and 10607 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/unit/conv/device/conv2d_testbed_interleaved.h
+4 -4
View File
@@ -257,15 +257,15 @@ public:
cutlass::conv::implicit_gemm_tensor_c_size(kConvolutionalOperator, problem_size),
{
reinterpret_cast<ElementAccumulator*> (workspace.get()),
ReductionStrideIndex(tensor_C.stride()[Conv2d::ImplicitGemmKernel::kTensorCStrideIdx])
ReductionStrideIndex(tensor_C.stride()[Conv2d::UnderlyingKernel::kTensorCStrideIdx])
},
{
tensor_D_computed.device_data(),
ReductionStrideIndex(tensor_C.stride()[Conv2d::ImplicitGemmKernel::kTensorCStrideIdx])
ReductionStrideIndex(tensor_C.stride()[Conv2d::UnderlyingKernel::kTensorCStrideIdx])
},
{
tensor_C.device_data(),
ReductionStrideIndex(tensor_C.stride()[Conv2d::ImplicitGemmKernel::kTensorCStrideIdx])
ReductionStrideIndex(tensor_C.stride()[Conv2d::UnderlyingKernel::kTensorCStrideIdx])
},
// apply alpha, beta to obtain the following equation alpha * ReduceAdd(A * B) + beta * C
{alpha, beta}
@@ -536,7 +536,7 @@ bool TestAllInterleavedConv2d(
// CUTLASS DGRAD's unity stride specialization only support stride {1, 1}
if ((ImplicitGemm::kConvolutionalOperator ==
cutlass::conv::Operator::kDgrad) &&
(ImplicitGemm::ImplicitGemmKernel::Mma::IteratorA::kStrideSupport ==
(ImplicitGemm::UnderlyingKernel::Mma::IteratorA::kStrideSupport ==
cutlass::conv::StrideSupport::kUnity)) {
if (!((conv_problem.stride_h == 1) && (conv_problem.stride_w == 1))) {
continue;