Conv3D Template Parameters

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

The following parameters need to be passed to create a Conv3D object:

Types of Input, Weight, Output, and Bias (optional) are defined by ConvType, including the logical memory location, data format, and data type.
Conv3dParam information (optional), which is used to enable the performance optimization template in different scenarios. Currently, this is not supported.

Prototype

template <class INPUT_TYPE, class WEIGHT_TYPE, class OUTPUT_TYPE, class BIAS_TYPE = biasType, class CONV_CFG = Conv3dParam>
using Conv3D = Conv3dIntfExt<Config<ConvApi::ConvDataType<INPUT_TYPE, WEIGHT_TYPE, OUTPUT_TYPE, BIAS_TYPE, CONV_CFG>>, Impl, Intf>

Parameters

**Table 1** Template parameters
Parameter	Input/Output	Description
INPUT_TYPE	Input	ConvType type template parameter, specifying the type of the Input parameter.
WEIGHT_TYPE	Input	ConvType type template parameter, specifying the type of the Weight parameter.
OUTPUT_TYPE	Input	ConvType type template parameter, specifying the type of the Output parameter.
BIAS_TYPE	Optional input	ConvType type template parameter, specifying the type of the Bias parameter.
CONV_CFG	Optional input	ConvParam type template parameter, which is used to enable the performance optimization template for different scenarios. In the current version, only the basic template is supported, and performance optimization is not enabled.

Returns

None

Restrictions

None

Example

#include "lib/conv/conv3d/conv3d_api.h"

using inputType = ConvApi::ConvType<AscendC::TPosition::GM, ConvFormat::NDC1HWC0, bfloat16_t>;
using weightType = ConvApi::ConvType<AscendC::TPosition::GM, ConvFormat::FRACTAL_Z_3D, bfloat16_t>;
using outputType = ConvApi::ConvType<AscendC::TPosition::GM, ConvFormat::NDC1HWC0, bfloat16_t>;
using biasType = ConvApi::ConvType<AscendC::TPosition::GM, ConvFormat::ND, float>; // The parameter is optional. If Bias is not used, the parameter does not need to be transferred.
struct ConvCustom : public ConvApi::ConvParam {
    __aicore__ inline ConvCustom(){};
}; // The parameter is optional. In the current version, only the basic template is supported, and performance optimization is not enabled. The parameter does not need to be transferred.

Conv3dApi::Conv3D<inputType, weightType, outputType, biasType, ConvCustom> conv3dApi;

Parent topic: Conv3D Kernel APIs