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.
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
update with onnx 1.11 release #10441
Conversation
#include "core/framework/TensorSeq.h" | ||
#include "core/providers/common.h" | ||
#include "core/framework/copy.h" | ||
#include "core/providers/op_kernel_type_control.h" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
are "core/framework/element_type_lists.h", "core/framework/TensorSeq.h", and "core/providers/op_kernel_type_control.h" needed?
this file doesn't seem to use any of the op kernel type control infrastructure or TensorSeq
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
can they be removed?
struct Func_Add<BFloat16> { | ||
void operator()(BFloat16*, const BFloat16*) const { | ||
ORT_NOT_IMPLEMENTED("CPU execution provider: BFloat16 data type is not supported with ScatterElements opset 16 when reduction is 'add'."); | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
general nit: run formatter on changed files
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
there's still some odd formatting here
@@ -120,7 +145,7 @@ Status ScatterNDBase::PrepareForCompute(OpKernelContext* context, Prepare& p) co | |||
auto* dst = output_tensor->template MutableData<std::string>(); | |||
std::copy(str_begin, str_end, dst); | |||
} else { | |||
memcpy(dst_base, src_base, input_tensor->SizeInBytes()); | |||
memcpy((void*)dst_base, (const void*)src_base, input_tensor->SizeInBytes()); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nit: casting to void* shouldn't be necessary
also prefer to avoid C-style casts, see https://google.github.io/styleguide/cppguide.html#Casting
thanks for updating this! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
signing off on the kernel def hash changes. not as familiar with the rest so it would be good if someone else can take a look too.
This PR is already in rel-1.11.0 branch, so remove the label. |
Description:
Update with onnx 1.11.0 release,
Fix build breaks due to new onnx 1.11.0 APIs,
Use FunctionBuilder from onnx,
support new and updated onnx ops with new and updated cpu kernels: ROIAlign ScatterElements, ScatterND, GridSample
Motivation and Context
prepare for onnxruntime 1.11.0 release