Fix SWDEV-483388: mistakenly invoke hipMemcpy when ORT_ENABLE_STREAM=…

…True
ROCm · Oct 16, 2024 · dc4c2a8 · dc4c2a8
1 parent 4573740
commit dc4c2a8
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/onnxruntime/core/providers/migraphx/gpu_data_transfer.cc b/onnxruntime/core/providers/migraphx/gpu_data_transfer.cc
@@ -57,7 +57,7 @@ common::Status GPUDataTransfer::CopyTensorAsync(const Tensor& src, Tensor& dst,
       HIP_CALL_THROW(hipMemcpyAsync(dst_data, src_data, bytes, hipMemcpyDeviceToDevice, static_cast<hipStream_t>(stream.GetHandle())));
     } else {
       // copy from other CPU memory to GPU, this is blocking
-      HIP_CALL_THROW(hipMemcpy(dst_data, src_data, bytes, hipMemcpyHostToDevice));
+      HIP_CALL_THROW(hipMemcpyWithStream(dst_data, src_data, bytes, hipMemcpyHostToDevice, static_cast<hipStream_t>(stream.GetHandle())));
     }
   } else if (src_device.Type() == OrtDevice::GPU) {
 #ifndef MIGRAPHX_STREAM_SYNC
@@ -66,7 +66,7 @@ common::Status GPUDataTransfer::CopyTensorAsync(const Tensor& src, Tensor& dst,
       HIP_CALL_THROW(hipMemcpyAsync(dst_data, src_data, bytes, hipMemcpyDeviceToHost, static_cast<hipStream_t>(stream.GetHandle())));
     } else {
       // copying from GPU to CPU memory, this is blocking
-      HIP_CALL_THROW(hipMemcpy(dst_data, src_data, bytes, hipMemcpyDeviceToHost));
+      HIP_CALL_THROW(hipMemcpyWithStream(dst_data, src_data, bytes, hipMemcpyDeviceToHost, static_cast<hipStream_t>(stream.GetHandle())));
     }
 #else
     HIP_CALL_THROW(hipMemcpyAsync(dst_data, src_data, bytes, hipMemcpyDeviceToHost, static_cast<hipStream_t>(stream.GetHandle())));