SetSelfDefineData

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 required computation data or data address stored in GM for use by the callback function when the template parameter MatmulCallBackFunc (custom callback function) is enabled. If the same Matmul object is reused, this API can be called multiple times to reset the corresponding data.

Prototype

__aicore__ inline void SetSelfDefineData(const uint64_t dataPtr)

__aicore__ inline void SetSelfDefineData(T dataPtr)

For Atlas A3 training products/Atlas A3 inference products, the SetSelfDefineData(T dataPtr) API prototype is not supported.

For Atlas A2 training products/Atlas A2 inference products, the SetSelfDefineData(T dataPtr) API prototype is not supported.

Parameters

**Table 1** Parameters
Parameter	Input/Output	Description
dataPtr	Input	Computation data or the data address stored in GM required by the operator's callback function. The type T supports user-defined basic structures.

Returns

None

Restrictions

If dataPtr is required in the callback function, this API must be called. If it is not required, this API does not need to be called.
This API is not supported when enableMixDualMaster (dual-master mode) is set to true.
This API must be called before the SetTensorA and SetTensorB APIs.

Example

// User-defined callback function
void DataCopyOut(const __gm__ void *gm, const LocalTensor<int8_t> &co1Local, const void *dataCopyOutParams, const uint64_t tilingPtr, const uint64_t dataPtr);
void CopyA1(const LocalTensor<int8_t> &aMatrix, const __gm__ void *gm, int row, int col, int useM, int useK, const uint64_t tilingPtr, const uint64_t dataPtr);
void CopyB1(const LocalTensor<int8_t> &bMatrix, const __gm__ void *gm, int row, int col, int useK, int useN, const uint64_t tilingPtr, const uint64_t dataPtr);

typedef AscendC::MatmulType<AscendC::TPosition::GM, CubeFormat::ND, half> aType; 
typedef AscendC::MatmulType<AscendC::TPosition::GM, CubeFormat::ND, half> bType; 
typedef AscendC::MatmulType<AscendC::TPosition::GM, CubeFormat::ND, float> cType; 
typedef AscendC::MatmulType<AscendC::TPosition::GM, CubeFormat::ND, float> biasType; 
AscendC::Matmul<aType, bType, cType, biasType, CFG_NORM, MatmulCallBackFunc<DataCopyOut, CopyA1, CopyB1>> mm;
REGIST_MATMUL_OBJ(&pipe, GetSysWorkSpacePtr(), mm, &tiling);
GlobalTensor<SrcT> dataGM; // Store the GM of the computation data required by the callback function.
uint64_t dataGMPtr = reinterpret_cast<uint64_t>(dataGM.address_);
mm.SetSelfDefineData(dataGMPtr);
mm.SetTensorA(gmA);
mm.SetTensorB(gmB);
mm.IterateAll();

Parent topic: Matmul Kernel APIs