Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Updated version of cuTENSOR and fixed paths #166

Merged
merged 1 commit into from
Apr 22, 2022
Merged

Updated version of cuTENSOR and fixed paths #166

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion cmake/FindcuTENSOR.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -79,7 +79,7 @@ else()
endif()

if(NOT cuTENSOR_FOUND)
set(CUTENSOR_VERSION 1.4.0.6)
set(CUTENSOR_VERSION 1.5.0.3)
set(CUTENSOR_FILENAME libcutensor-linux-x86_64-${CUTENSOR_VERSION}-archive)

message(STATUS "cuTENSOR not found. Downloading library. By continuing this download you accept to the license terms of cuTENSOR")
Expand Down
6 changes: 3 additions & 3 deletions cmake/FindcuTensorNet.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,10 +81,10 @@ endif()
if(NOT cuTensorNet_FOUND)
message(STATUS "cuTensorNet not found. Downloading library. By continuing this download you accept to the license terms of cuQuantum SDK")

set(CUTENSORNET_VERSION 0.1.0.30)
set(CUTENSORNET_VERSION 22.03.0.40)
set(CUTENSORNET_FILENAME cuquantum-linux-x86_64-${CUTENSORNET_VERSION}-archive)
file(DOWNLOAD https://developer.download.nvidia.com/compute/cuquantum/redist/linux-x86_64/${CUTENSORNET_FILENAME}.tar.xz

file(DOWNLOAD https://developer.download.nvidia.com/compute/cuquantum/redist/cuquantum/linux-x86_64/${CUTENSORNET_FILENAME}.tar.xz
${CMAKE_BINARY_DIR}/${CUTENSORNET_FILENAME}.tar.xz)

file(ARCHIVE_EXTRACT INPUT ${CMAKE_BINARY_DIR}/${CUTENSORNET_FILENAME}.tar.xz DESTINATION ${CMAKE_BINARY_DIR}/cutensornet/)
Expand Down
48 changes: 37 additions & 11 deletions include/matx_einsum.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -127,6 +127,15 @@ class matxEinsumHandle_t {
MATX_ASSERT_STR(cutensornetCreateContractionOptimizerInfo(handle_, descNet_, &optimizerInfo) == CUTENSORNET_STATUS_SUCCESS, matxcuTensorError,
"Failed to create cuTensorNet contraction optimizer info");

int imbalance_factor = 30;
MATX_ASSERT_STR(cutensornetContractionOptimizerConfigSetAttribute(
handle_,
optimizerConfig,
CUTENSORNET_CONTRACTION_OPTIMIZER_CONFIG_GRAPH_IMBALANCE_FACTOR,
&imbalance_factor,
sizeof(imbalance_factor)) == CUTENSORNET_STATUS_SUCCESS,
matxcuTensorError, "Failed to run contraction optimizer");

size_t freeMem, totalMem;
MATX_ASSERT(cudaMemGetInfo(&freeMem, &totalMem) == cudaSuccess, matxCudaError);

Expand All @@ -150,13 +159,37 @@ class matxEinsumHandle_t {

MATX_ASSERT(params_.num_slices_ > 0, matxcuTensorError);

MATX_ASSERT(cutensornetCreateWorkspaceDescriptor(handle_, &workDesc_) == CUTENSORNET_STATUS_SUCCESS, matxcuTensorError);

uint64_t requiredWorkspaceSize = 0;
MATX_ASSERT(cutensornetWorkspaceComputeSizes(handle_,
descNet_,
optimizerInfo,
workDesc_) == CUTENSORNET_STATUS_SUCCESS, matxcuTensorError);

MATX_ASSERT(cutensornetWorkspaceGetSize(handle_,
workDesc_,
CUTENSORNET_WORKSIZE_PREF_MIN,
CUTENSORNET_MEMSPACE_DEVICE,
&requiredWorkspaceSize) == CUTENSORNET_STATUS_SUCCESS, matxcuTensorError);

MATX_ASSERT_STR(workSize_ > requiredWorkspaceSize, matxOutOfMemory, "Not enough workspace memory is available.");

matxAlloc(&workspace_, workSize_, MATX_ASYNC_DEVICE_MEMORY, stream);

MATX_ASSERT (cutensornetWorkspaceSet(handle_,
workDesc_,
CUTENSORNET_MEMSPACE_DEVICE,
workspace_,
workSize_) == CUTENSORNET_STATUS_SUCCESS, matxcuTensorError);

/*******************************
* Initialize all pair-wise contraction plans (for cuTENSOR)
*******************************/
MATX_ASSERT_STR(cutensornetCreateContractionPlan(handle_,
descNet_,
optimizerInfo,
workSize_,
workDesc_,
&plan_) == CUTENSORNET_STATUS_SUCCESS,
matxcuTensorError, "cutensornetCreateContractionPlan failed");

Expand All @@ -169,12 +202,6 @@ class matxEinsumHandle_t {
&autotunePref) == CUTENSORNET_STATUS_SUCCESS,
matxcuTensorError, "cutensornetCreateContractionAutotunePreference failed");

// Allocate the real amount needed and free the old amount
MATX_ASSERT_STR(cutensornetContractionGetWorkspaceSize(handle_, descNet_, optimizerInfo, &workSize_) == CUTENSORNET_STATUS_SUCCESS,
matxcuTensorError, "cutensornetContractionGetWorkspaceSize failed");

matxAlloc(&workspace_, workSize_, MATX_ASYNC_DEVICE_MEMORY, stream);

const int numAutotuningIterations = 5; // may be 0
MATX_ASSERT_STR(cutensornetContractionAutotunePreferenceSetAttribute(
handle_,
Expand All @@ -189,8 +216,7 @@ class matxEinsumHandle_t {
plan_,
data_in,
out.Data(),
workspace_,
workSize_,
workDesc_,
autotunePref,
stream) == CUTENSORNET_STATUS_SUCCESS,
matxcuTensorError, "cutensornetContractionAutotune failed");
Expand Down Expand Up @@ -329,8 +355,7 @@ class matxEinsumHandle_t {
plan_,
data_in,
out.Data(),
workspace_,
workSize_,
workDesc_,
slice,
stream) == CUTENSORNET_STATUS_SUCCESS,
matxcuTensorError, "cutensornetContraction failed");
Expand All @@ -348,6 +373,7 @@ class matxEinsumHandle_t {
cutensornetContractionPlan_t plan_;
uint64_t workSize_;
void *workspace_;
cutensornetWorkspaceDescriptor_t workDesc_;
cutensornetHandle_t handle_;
cutensornetNetworkDescriptor_t descNet_;
EinsumParams_t<InT...> params_;
Expand Down
2 changes: 1 addition & 1 deletion test/00_tensor/EinsumTests.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -98,7 +98,7 @@ TYPED_TEST_SUITE(EinsumTestsIntegral, MatXAllIntegralTypes);
TYPED_TEST_SUITE(EinsumTestsNumericNonComplex, MatXNumericNonComplexTypes);
TYPED_TEST_SUITE(EinsumTestsBoolean, MatXBoolTypes);

#if ENABLE_CUTENSOR
#if MATX_ENABLE_CUTENSOR
TYPED_TEST(EinsumTestsFloatNonComplexNonHalfTypes, Contraction3D)
{
MATX_ENTER_HANDLER();
Expand Down