BatchMatMul Class Reference

#include <BatchMatMulImpl.hpp>

Public Member Functions
	BatchMatMul (const BatchMatMulDescriptor &params, const TensorInfo &inputXInfo, const TensorInfo &inputYInfo, const TensorInfo &outputInfo, Decoder< float > &inputXDecoder, Decoder< float > &inputYDecoder, Encoder< float > &outputEncoder)

Detailed Description

Definition at line 16 of file BatchMatMulImpl.hpp.

Constructor & Destructor Documentation

BatchMatMul()

BatchMatMul	(	const BatchMatMulDescriptor &	params,
		const TensorInfo &	inputXInfo,
		const TensorInfo &	inputYInfo,
		const TensorInfo &	outputInfo,
		Decoder< float > &	inputXDecoder,
		Decoder< float > &	inputYDecoder,
		Encoder< float > &	outputEncoder
	)

Definition at line 15 of file BatchMatMulImpl.cpp.

References ARMNN_ASSERT, Decoder< IType >::DecodeTensor(), Encoder< IType >::Get(), BatchMatMulDescriptor::GetAxesToMul(), TensorInfo::GetNumDimensions(), BatchMatMulDescriptor::GetPermuteVec(), TensorInfo::GetShape(), BatchMatMulDescriptor::m_AdjointX, BatchMatMulDescriptor::m_AdjointY, BatchMatMulDescriptor::m_DataLayoutX, BatchMatMulDescriptor::m_DataLayoutY, BatchMatMulDescriptor::m_TransposeX, BatchMatMulDescriptor::m_TransposeY, armnn::Output, armnn::Permute, armnnUtils::Permuted(), and Encoder< IType >::Set().

    : params(params),
      inputXInfo(inputXInfo),
      inputYInfo(inputYInfo),
      outputInfo(outputInfo),
      inputXDecoder(inputXDecoder),
      inputYDecoder(inputYDecoder),
      outputEncoder(outputEncoder)
{
    inputXData = this->inputXDecoder.DecodeTensor(inputXInfo.GetShape());
    inputYData = this->inputYDecoder.DecodeTensor(inputYInfo.GetShape());
    // At this point, we don't touch the input decoders - just the resultant vectors

    ApplyParams();

    ApplyBatchMatMul();
}

---

The documentation for this class was generated from the following files:

• src/backends/reference/workloads/BatchMatMulImpl.hpp
• src/backends/reference/workloads/BatchMatMulImpl.cpp