SetGradOutput

Applicability

Product	Supported
Atlas A3 training products/Atlas A3 inference products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	x
Atlas inference product's AI Core	x
Atlas inference product's Vector Core	x
Atlas training products	x

Function

Sets the input matrix GradOutput for convolution backpropagation computation.

Prototype

__aicore__ inline void SetGradOutput(const AscendC::GlobalTensor<SrcT> &gradOutput)

Parameters

**Table 1** API parameters
Parameter	Input/Output	Description
gradOutput	Input	Start address of the GradOutput matrix in the global memory. The type is GlobalTensor. SrcT indicates the data type of the GradOutput matrix. Currently, the supported data types are half and bfloat16_t.

Returns

None

Restrictions

None

Example

const Conv3DBackpropFilterTilingData* tilingData;
//... Initialize tilingData.
ConvBackpropApi::Conv3DBackpropFilter <inputType, weightSizeType, gradOutputType, gradWeightType> gradWeight_;
gradWeight_.Init(&(tilingData->dwTiling));
gradWeight_.SetInput(inputGm_[offsetB_]);
// Set gradOutput.
gradWeight_.SetGradOutput(gradOutputGm_[offsetA_]);
...  

Parent topic: Conv3DBackpropFilter Kernel API