|
|
|
@ -190,6 +190,7 @@ void PyCUDATensorSetFromArray(
|
|
|
|
|
static_cast<const platform::CUDADeviceContext *>(pool.Get(place));
|
|
|
|
|
paddle::platform::GpuMemcpyAsync(dst, array.data(), sizeof(T) * array.size(),
|
|
|
|
|
cudaMemcpyHostToDevice, dev_ctx->stream());
|
|
|
|
|
dev_ctx->Wait();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
template <>
|
|
|
|
|