saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-20 19:12:24 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Conv node bug, cached state was incoherent (#10041)

Browse source 
* Moved the init earlier to keep the cache coherent
* Move setting of w_desc later, and zero shape check later to catch all cacheable changes.
* Add comment
This commit is contained in:
Ryan Hill 2022-03-01 01:31:57 -08:00 committed by GitHub
parent f4b2d3af2b
commit c1cf16ed5d
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 6 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
onnxruntime/core/providers/cuda/nn/conv.cc
View file

@ -176,9 +176,6 @@ Status Conv<T>::UpdateState(OpKernelContext* context, bool bias_expected) const
s_.slice_axes = slice_axes;
s_.Y = context->Output(0, TensorShape(s_.y_dims));
if (s_.Y->Shape().Size() == 0) {
return Status::OK();
}
if (post_slicing_required) {
// Post slicing needed. Create and fill in the Conv results in an intermediate buffer.
s_.memory_for_cudnn_conv_results = GetScratchBuffer<void>(TensorShape(y_dims_with_adjusted_pads).Size() * s_.element_size);
@ -206,9 +203,14 @@ Status Conv<T>::UpdateState(OpKernelContext* context, bool bias_expected) const
dilations.push_back(1);
}
if (w_dims_changed) {
if (w_dims_changed)
ORT_RETURN_IF_ERROR(s_.w_desc.Set(w_dims, CudnnTensor::GetDataType<CudaT>()));
// We must delay returning early until here so that the weight dims have been cached properly
if (s_.Y->Shape().Size() == 0) {
return Status::OK();
}
ORT_RETURN_IF_ERROR(s_.x_tensor.Set(x_dims_cudnn, CudnnTensor::GetDataType<CudaT>()));
ORT_RETURN_IF_ERROR(s_.y_tensor.Set(y_dims_cudnn, CudnnTensor::GetDataType<CudaT>()));
ORT_RETURN_IF_ERROR(s_.conv_desc.Set(kernel_shape.size(), pads, strides, dilations,