saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-08 17:17:15 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix the cuda EP test failure (#14087)

Browse source 
### Description
Fix a regression failure for cuda EP test



### Motivation and Context
CudaEP test is a special test case under EP folder, not in test folder.
when refactor the code during multi-stream work, we missed it. This PR
is to fix the test.

Co-authored-by: Cheng Tang <chenta@microsoft.com@orttrainingdev9.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
This commit is contained in:
Tang, Cheng 2022-12-28 17:04:27 -08:00 committed by GitHub
parent a79d88ed7f
commit 9f52a8bc55
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 8 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
onnxruntime/core/providers/cuda/test/cuda_execution_provider_test.cc
View file

@ -15,7 +15,8 @@
namespace onnxruntime {
namespace cuda {
namespace test {
// TODO: Since the "DeferredRelease" has been migrated to CudaStream class,
// we should migrate this test from CudaEP unit test to CudaStream unit test.
bool TestDeferredRelease() {
// Create CUDA EP.
CUDAExecutionProviderInfo info;
@ -27,7 +28,9 @@ bool TestDeferredRelease() {
// Allocator for call cudaMallocHost and cudaFreeHost
// For details, see CUDAPinnedAllocator in cuda_allocator.cc.
AllocatorPtr cpu_pinned_alloc = ep.GetAllocator(DEFAULT_CPU_ALLOCATOR_DEVICE_ID, OrtMemTypeCPU);
CudaStream stream(nullptr, gpu_alloctor->Info().device, cpu_pinned_alloc, false, false, nullptr, nullptr);
// let the CudaStream instance "own" the default stream, so we can avoid the
// work to initialize cublas/cudnn/... It is ok since it is just a customized unit test.
CudaStream stream(nullptr, gpu_alloctor->Info().device, cpu_pinned_alloc, false, true, nullptr, nullptr);
// 10 MB
const size_t n_bytes = 10 * 1000000;
const int64_t n_allocs = 64;
@ -72,7 +75,9 @@ bool TestDeferredReleaseWithoutArena() {
// Allocator for call cudaMallocHost and cudaFreeHost
// For details, see CUDAPinnedAllocator in cuda_allocator.cc.
AllocatorPtr cpu_pinned_alloc = ep.GetAllocator(DEFAULT_CPU_ALLOCATOR_DEVICE_ID, OrtMemTypeCPU);
CudaStream stream(nullptr, gpu_alloctor->Info().device, cpu_pinned_alloc, false, false, nullptr, nullptr);
// let the CudaStream instance "own" the default stream, so we can avoid the
// work to initialize cublas/cudnn/... It is ok since it is just a customized unit test.
CudaStream stream(nullptr, gpu_alloctor->Info().device, cpu_pinned_alloc, false, true, nullptr, nullptr);
// 10 MB
const size_t n_bytes = 10 * 1000000;
const int64_t n_allocs = 64;