/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck_tile/ops/batched_contraction/utils/tensor_descriptor_utils.hpp File Reference

• develop
• include
• ck_tile
• ops
• batched_contraction
• utils

Classes | Namespaces

tensor_descriptor_utils.hpp File Reference

Utility functions for creating tensor descriptors in batched contraction operations. More...

#include "ck_tile/core.hpp"

Go to the source code of this file.

Classes
struct	ck_tile::TensorDescriptorUtils< NumDimG, NumDimM, NumDimN, NumDimK, VectorSizeA, VectorSizeB, VectorSizeE >
	Utility class for creating tensor descriptors in batched contraction operations. More...

Namespaces
	ck_tile

Detailed Description

Utility functions for creating tensor descriptors in batched contraction operations.

This file contains utility functions for creating tensor descriptors with flattened dimensions for GEMM operations. These functions transform multi-dimensional tensors into 2D matrix descriptors by removing batch dimensions and flattening the remaining dimensions.

These utilities are used by BatchedContractionKernel to create stride-aware descriptors that support arbitrary multi-dimensional non-contiguous tensor layouts.