Skip to content

[Kernel] Fix einsum operator for empty inputs #27226

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.

Sign up for GitHub

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

Merged
hariharans29 merged 12 commits into from
Feb 5, 2026
Merged

[Kernel] Fix einsum operator for empty inputs #27226

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
6fb223f
Dupe of 23379
hariharans29 Feb 3, 2026
7e83f1d
Avoid unnecessary copy on CPU
hariharans29 Feb 3, 2026
c74b516
Update onnxruntime/core/providers/cpu/math/einsum_utils/einsum_typed_…
hariharans29 Feb 3, 2026
e0e6e01
More changes
hariharans29 Feb 3, 2026
9b9634f
Fixes
hariharans29 Feb 3, 2026
b845a7c
Adding infra for zeroing output buffer
hariharans29 Feb 3, 2026
987d647
Update onnxruntime/core/providers/cuda/math/einsum_utils/einsum_auxil…
hariharans29 Feb 3, 2026
1522f1d
Fixes
hariharans29 Feb 3, 2026
934cf05
Merge branch 'hari/einsum_fix_3' of https://github.com/microsoft/onnx…
hariharans29 Feb 3, 2026
f24813d
Merge remote-tracking branch 'origin' into hari/einsum_fix_3
hariharans29 Feb 4, 2026
edb352d
Consistency
hariharans29 Feb 5, 2026
4ef7a81
Rename Zeroing
hariharans29 Feb 5, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions onnxruntime/core/providers/cpu/cpu_provider_shared.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -186,9 +186,9 @@ struct ProviderHostCPUImpl : ProviderHostCPU {
std::unique_ptr<EinsumTypedComputeProcessor<float>> EinsumTypedComputeProcessor_float__Create(OpKernelContext* context, AllocatorPtr allocator, concurrency::ThreadPool* tp, EinsumComputePreprocessor& einsum_compute_preprocessor, void* einsum_cuda_assets) override { return std::make_unique<EinsumTypedComputeProcessor<float>>(context, allocator, tp, einsum_compute_preprocessor, einsum_cuda_assets); }
std::unique_ptr<EinsumTypedComputeProcessor<double>> EinsumTypedComputeProcessor_double__Create(OpKernelContext* context, AllocatorPtr allocator, concurrency::ThreadPool* tp, EinsumComputePreprocessor& einsum_compute_preprocessor, void* einsum_cuda_assets) override { return std::make_unique<EinsumTypedComputeProcessor<double>>(context, allocator, tp, einsum_compute_preprocessor, einsum_cuda_assets); }
std::unique_ptr<EinsumTypedComputeProcessor<MLFloat16>> EinsumTypedComputeProcessor_MLFloat16__Create(OpKernelContext* context, AllocatorPtr allocator, concurrency::ThreadPool* tp, EinsumComputePreprocessor& einsum_compute_preprocessor, void* einsum_cuda_assets) override { return std::make_unique<EinsumTypedComputeProcessor<MLFloat16>>(context, allocator, tp, einsum_compute_preprocessor, einsum_cuda_assets); }
void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<float>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<float>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<float>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) override { return p->SetDeviceHelpers(device_transpose_func, device_matmul_func, device_reduce_sum_func, device_data_copy_func); }
void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<double>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<double>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<double>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) override { return p->SetDeviceHelpers(device_transpose_func, device_matmul_func, device_reduce_sum_func, device_data_copy_func); }
void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<MLFloat16>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<MLFloat16>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<MLFloat16>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) override { return p->SetDeviceHelpers(device_transpose_func, device_matmul_func, device_reduce_sum_func, device_data_copy_func); }
void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<float>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<float>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<float>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func, const EinsumOp::DeviceHelpers::Zeroing& device_zeroing_func) override { return p->SetDeviceHelpers(device_transpose_func, device_matmul_func, device_reduce_sum_func, device_data_copy_func, device_zeroing_func); }
void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<double>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<double>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<double>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func, const EinsumOp::DeviceHelpers::Zeroing& device_zeroing_func) override { return p->SetDeviceHelpers(device_transpose_func, device_matmul_func, device_reduce_sum_func, device_data_copy_func, device_zeroing_func); }
void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<MLFloat16>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<MLFloat16>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<MLFloat16>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func, const EinsumOp::DeviceHelpers::Zeroing& device_zeroing_func) override { return p->SetDeviceHelpers(device_transpose_func, device_matmul_func, device_reduce_sum_func, device_data_copy_func, device_zeroing_func); }
Status EinsumTypedComputeProcessor__Run(EinsumTypedComputeProcessor<float>* p) override { return p->Run(); }
Status EinsumTypedComputeProcessor__Run(EinsumTypedComputeProcessor<double>* p) override { return p->Run(); }
Status EinsumTypedComputeProcessor__Run(EinsumTypedComputeProcessor<MLFloat16>* p) override { return p->Run(); }
Expand Down
6 changes: 3 additions & 3 deletions onnxruntime/core/providers/cpu/cpu_provider_shared.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -117,9 +117,9 @@ struct ProviderHostCPU {
virtual std::unique_ptr<EinsumTypedComputeProcessor<float>> EinsumTypedComputeProcessor_float__Create(OpKernelContext* context, AllocatorPtr allocator, concurrency::ThreadPool* tp, EinsumComputePreprocessor& einsum_compute_preprocessor, void* einsum_cuda_assets) = 0;
virtual std::unique_ptr<EinsumTypedComputeProcessor<double>> EinsumTypedComputeProcessor_double__Create(OpKernelContext* context, AllocatorPtr allocator, concurrency::ThreadPool* tp, EinsumComputePreprocessor& einsum_compute_preprocessor, void* einsum_cuda_assets) = 0;
virtual std::unique_ptr<EinsumTypedComputeProcessor<MLFloat16>> EinsumTypedComputeProcessor_MLFloat16__Create(OpKernelContext* context, AllocatorPtr allocator, concurrency::ThreadPool* tp, EinsumComputePreprocessor& einsum_compute_preprocessor, void* einsum_cuda_assets) = 0;
virtual void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<float>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<float>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<float>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) = 0;
virtual void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<double>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<double>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<double>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) = 0;
virtual void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<MLFloat16>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<MLFloat16>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<MLFloat16>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) = 0;
virtual void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<float>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<float>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<float>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func, const EinsumOp::DeviceHelpers::Zeroing& device_zeroing_func) = 0;
virtual void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<double>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<double>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<double>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func, const EinsumOp::DeviceHelpers::Zeroing& device_zeroing_func) = 0;
virtual void EinsumTypedComputeProcessor__SetDeviceHelpers(EinsumTypedComputeProcessor<MLFloat16>* p, const EinsumOp::DeviceHelpers::Transpose& device_transpose_func, const EinsumOp::DeviceHelpers::MatMul<MLFloat16>& device_matmul_func, const EinsumOp::DeviceHelpers::ReduceSum<MLFloat16>& device_reduce_sum_func, const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func, const EinsumOp::DeviceHelpers::Zeroing& device_zeroing_func) = 0;
virtual Status EinsumTypedComputeProcessor__Run(EinsumTypedComputeProcessor<float>* p) = 0;
virtual Status EinsumTypedComputeProcessor__Run(EinsumTypedComputeProcessor<double>* p) = 0;
virtual Status EinsumTypedComputeProcessor__Run(EinsumTypedComputeProcessor<MLFloat16>* p) = 0;
Expand Down
12 changes: 8 additions & 4 deletions onnxruntime/core/providers/cpu/math/einsum.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,7 +65,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor.SetDeviceHelpers(EinsumOp::DeviceHelpers::CpuDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::MatMul<float>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::ReduceSum<float>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::Zeroing);
return einsum_compute_processor.Run();
} else if (inputs[0]->IsDataType<int32_t>()) {
auto einsum_compute_processor = EinsumTypedComputeProcessor<int32_t>(context,
Expand All @@ -78,7 +79,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor.SetDeviceHelpers(EinsumOp::DeviceHelpers::CpuDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::MatMul<int32_t>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::ReduceSum<int32_t>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::Zeroing);

return einsum_compute_processor.Run();
} else if (inputs[0]->IsDataType<double>()) {
Expand All @@ -92,7 +94,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor.SetDeviceHelpers(EinsumOp::DeviceHelpers::CpuDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::MatMul<double>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::ReduceSum<double>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::Zeroing);
return einsum_compute_processor.Run();
} else if (inputs[0]->IsDataType<int64_t>()) {
auto einsum_compute_processor = EinsumTypedComputeProcessor<int64_t>(context,
Expand All @@ -104,7 +107,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor.SetDeviceHelpers(EinsumOp::DeviceHelpers::CpuDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::MatMul<int64_t>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::ReduceSum<int64_t>,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CpuDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CpuDeviceHelpers::Zeroing);

return einsum_compute_processor.Run();
}
Expand Down
6 changes: 6 additions & 0 deletions onnxruntime/core/providers/cpu/math/einsum_utils/einsum_auxiliary_ops.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,12 @@ Status DataCopy(const Tensor& input, Tensor& output, void* /*einsum_cuda_assets*
return Status::OK();
}

// CPU specific Zeroing helper
Status Zeroing(Tensor& input, void* /*einsum_cuda_assets*/) {
memset(input.MutableDataRaw(), 0, input.SizeInBytes());
return Status::OK();
}

// CPU specific Transpose helper
Status Transpose(const gsl::span<const size_t>& permutation, const Tensor& input,
Tensor& output, const TensorShape* input_shape_override, void* /*einsum_cuda_assets*/) {
Expand Down
5 changes: 5 additions & 0 deletions onnxruntime/core/providers/cpu/math/einsum_utils/einsum_auxiliary_ops.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,9 @@ namespace DeviceHelpers {
// Data copy op - Copies raw data from the source tensor's buffer to the destination tensor's buffer
using DataCopy = std::function<Status(const Tensor& input, Tensor& output, void* einsum_cuda_assets)>;

// Zeroing op - Sets all bytes in the tensor's buffer to zero
using Zeroing = std::function<Status(Tensor& input, void* einsum_cuda_assets)>;

// Transpose op - Transposes given input based on data in `permutation`
using Transpose = std::function<Status(const gsl::span<const size_t>& permutation, const Tensor& input,
Tensor& output, const TensorShape* input_shape_override,
Expand Down Expand Up @@ -63,6 +66,8 @@ namespace CpuDeviceHelpers {

Status DataCopy(const Tensor& input, Tensor& output, void* einsum_cuda_assets);

Status Zeroing(Tensor& input, void* einsum_cuda_assets);

Status Transpose(const gsl::span<const size_t>& permutation, const Tensor& input,
Tensor& output, const TensorShape* input_shape_override, void* einsum_cuda_assets);

Expand Down
18 changes: 17 additions & 1 deletion onnxruntime/core/providers/cpu/math/einsum_utils/einsum_typed_compute_processor.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -336,11 +336,13 @@ template <typename T>
void EinsumTypedComputeProcessor<T>::SetDeviceHelpers(const EinsumOp::DeviceHelpers::Transpose& device_transpose_func,
const EinsumOp::DeviceHelpers::MatMul<T>& device_matmul_func,
const EinsumOp::DeviceHelpers::ReduceSum<T>& device_reduce_sum_func,
const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func) {
const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func,
const EinsumOp::DeviceHelpers::Zeroing& zero_input_buffer_func) {
device_transpose_func_ = device_transpose_func;
device_matmul_func_ = device_matmul_func;
device_reduce_sum_func_ = device_reduce_sum_func;
device_data_copy_func_ = device_data_copy_func;
zero_input_buffer_func_ = zero_input_buffer_func;
}

template <typename T>
Expand All @@ -357,6 +359,20 @@ Status EinsumTypedComputeProcessor<T>::Run() {

auto num_inputs = context_->InputCount();

{
bool has_empty_input = std::any_of(raw_inputs.begin(), raw_inputs.end(), [](const auto& input) {
return input->Shape().Size() == 0;
});

// Skip all the work, fill with zeros if needed
if (has_empty_input) {
const auto output_dims = einsum_compute_preprocessor_.GetOutputDims();
Tensor& output = *context_->Output(0, output_dims);

return zero_input_buffer_func_(output, einsum_ep_assets_);
}
}

// Pre-process the first input so as to reduce any dims that only it has
std::unique_ptr<const Tensor> result;

Expand Down
4 changes: 3 additions & 1 deletion onnxruntime/core/providers/cpu/math/einsum_utils/einsum_typed_compute_processor.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,8 @@ class EinsumTypedComputeProcessor {
void SetDeviceHelpers(const EinsumOp::DeviceHelpers::Transpose& device_transpose_func,
const EinsumOp::DeviceHelpers::MatMul<T>& device_matmul_func,
const EinsumOp::DeviceHelpers::ReduceSum<T>& device_reduce_sum_func,
const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func);
const EinsumOp::DeviceHelpers::DataCopy& device_data_copy_func,
const EinsumOp::DeviceHelpers::Zeroing& zero_input_buffer_func);

Status Run();

Expand Down Expand Up @@ -64,6 +65,7 @@ class EinsumTypedComputeProcessor {
EinsumOp::DeviceHelpers::MatMul<T> device_matmul_func_;
EinsumOp::DeviceHelpers::ReduceSum<T> device_reduce_sum_func_;
EinsumOp::DeviceHelpers::DataCopy device_data_copy_func_;
EinsumOp::DeviceHelpers::Zeroing zero_input_buffer_func_;

// Holds EP-specific assets required for (auxiliary) ops that need to be executed on non-CPU EPs
void* einsum_ep_assets_;
Expand Down
9 changes: 6 additions & 3 deletions onnxruntime/core/providers/cuda/math/einsum.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor->SetDeviceHelpers(EinsumOp::DeviceHelpers::CudaDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::MatMul<float>,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::ReduceSum<float>,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CudaDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::Zeroing);
return einsum_compute_processor->Run();
} else if (inputs[0]->IsDataType<double>()) {
auto einsum_compute_processor = EinsumTypedComputeProcessor<double>::Create(context, allocator, tp,
Expand All @@ -63,7 +64,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor->SetDeviceHelpers(EinsumOp::DeviceHelpers::CudaDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::MatMul<double>,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::ReduceSum<double>,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CudaDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::Zeroing);
return einsum_compute_processor->Run();
} else if (inputs[0]->IsDataType<MLFloat16>()) {
auto einsum_compute_processor = EinsumTypedComputeProcessor<MLFloat16>::Create(context, allocator, tp,
Expand All @@ -73,7 +75,8 @@ Status Einsum::DeviceCompute(OpKernelContext* context, const std::vector<const T
einsum_compute_processor->SetDeviceHelpers(EinsumOp::DeviceHelpers::CudaDeviceHelpers::Transpose,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::MatMul<MLFloat16>,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::ReduceSum<MLFloat16>,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::DataCopy);
EinsumOp::DeviceHelpers::CudaDeviceHelpers::DataCopy,
EinsumOp::DeviceHelpers::CudaDeviceHelpers::Zeroing);
return einsum_compute_processor->Run();
}

Expand Down
8 changes: 8 additions & 0 deletions onnxruntime/core/providers/cuda/math/einsum_utils/einsum_auxiliary_ops.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,14 @@ Status DataCopy(const Tensor& input, Tensor& output, void* einsum_cuda_assets) {
return Status::OK();
}


// CUDA EP specific Zeroing helper
Status Zeroing(Tensor& input, void* einsum_cuda_assets) {
CUDA_RETURN_IF_ERROR(cudaMemsetAsync(input.MutableDataRaw(), 0, input.SizeInBytes(),
static_cast<EinsumCudaAssets*>(einsum_cuda_assets)->GetCudaStream()));
return Status::OK();
}

// CUDA EP specific Transpose helper
Status Transpose(const gsl::span<const size_t>& permutation, const Tensor& input,
Tensor& output, const TensorShape* input_shape_override, void* einsum_cuda_assets) {
Expand Down
2 changes: 2 additions & 0 deletions onnxruntime/core/providers/cuda/math/einsum_utils/einsum_auxiliary_ops.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,6 +47,8 @@ Status Transpose(const gsl::span<const size_t>& permutation, const Tensor& input

Status DataCopy(const Tensor& input, Tensor& output, void* einsum_cuda_assets);

Status Zeroing(Tensor& input, void* einsum_cuda_assets);

template <typename T>
Status MatMul(const T* input_1_data, const T* input_2_data, T* output_data,
size_t left_stride, size_t right_stride, size_t output_stride,
Expand Down
Loading
Loading