Strmm

Applicable Products

Function Description

API function

• asdBlasMakeStrmmPlan: initializes the Strmm operator configuration corresponding to the handle.

• asdBlasStrmm: single precision. It is used to multiply a triangular matrix A by a matrix B to obtain a new matrix C.

Formula:

• asdBlasStrmm calculation formula:

  

• Example:

  Input A:

  [   [ 1, 0 ],
      [ 3, 4 ]  ]

  Input B:

  [   [ 1, 2 ],
      [ 3, 4 ]  ]

  The input side is L, the input uplo is L, the input trans is N, and the input diag is N.

  The input n is 2, the input lda is 2, the input ldb is 2, and the input ldc is 2.

  The input alpha is 2.345.

  After the Strmm operator is called, the output C is as follows:

  [  [ 2.3450,  4.6900],
     [35.1750, 51.5900]  ]

Function Prototype

• AspbStatus asdBlasMakeStrmmPlan(asdBlasHandle handle)
• AspbStatus asdBlasStrmm(asdBlasHandle handle, asdBlasSideMode_t side, asdBlasFillMode_t uplo, asdBlasOperation_t trans,asdBlasDiagType_t diag, const int64_t m, const int64_t n, const float *alpha, aclTensor *A,const int64_t lda, aclTensor *B, const int64_t ldb, aclTensor *C, const int64_t ldc)

Parameter Description

• asdBlasMakeStrmmPlan

  Parameter	Input/Output	Type	Description
  handle	Input	asdBlasHandle	Handle of the Strmm operator.

• asdBlasStrmm

  Parameter	Input/Output	Type	Description
  handle	Input	asdBlasHandle	Handle of the Strmm operator.
  side	Input	asdBlasSideMode_t	Specifies whether matrix A is on the left or right side of the multiplication. ASDBLAS_SIDE_LEFT // Left ASDBLAS_SIDE_RIGHT // Right
  uplo	Input	asdBlasFillMode_t	Storage mode of matrix A. ASDBLAS_FILL_MODE_LOWER // Lower triangle ASDBLAS_FILL_MODE_UPPER // Upper triangle
  trans	Input	asdBlasOperation_t	Whether to transpose matrix A. ASDBLAS_OP_N // Not transposed ASDBLAS_OP_T // Transposed
  diag	Input	asdBlasDiagType_t	Specifies whether the diagonal elements of matrix A are assumed to be 1. ASDBLAS_DIAG_NON_UNIT // Not assumed to be 1 ASDBLAS_DIAG_UNIT // Assumed to be 1
  m	Input	const int64_t	Number of rows in matrix B and matrix C.
  n	Input	const int64_t	Number of columns in matrix B and matrix C.
  alpha	Input	const float *	alpha in the formula, which the coefficient used to compute the matrix multiplication.
  A	Input	aclTensor *	A in the formula, which is a tensor on the device. The data type can only be FLOAT32, the data format can be ND, and the shape is [n, n]. When A is the left-side multiplication matrix, the shape is [m, m]; when A is the right-side multiplication matrix, the shape is [n, n].
  lda	Input	const int64_t	Stride of elements in tensor A (currently constrained to m/n: m when side=ASDBLAS_SIDE_LEFT, and n when side=ASDBLAS_SIDE_RIGHT).
  B	Input	aclTensor *	B in the formula, which is a tensor on the device. The data type can only be FLOAT32, the data format can be ND, and the shape is [m, n].
  ldb	Input	const int64_t	Stride of elements in tensor B (currently constrained to m).
  C	Output	aclTensor *	C in the formula, which is a tensor on the device. The data type can only be FLOAT32, the data format can be ND, and the shape is [m, n].
  ldc	Input	const int64_t	Stride of elements in tensor C (currently constrained to m).

Return Value Description

For details about the return values, see Return Value.

Constraints

• asdBlasMakeStrmmPlan: none.

• asdBlasStrmm
  • The input element count (m, n) currently supports the range [1, 8193].
  • When side is set to ASDBLAS_SIDE_LEFT, the input shape of the operator is [m, m] and [m, n], and the output shape is [m, n].
  • When side is set to ASDBLAS_SIDE_RIGHT, the input shape of the operator is [n, n] and [m, n], and the output shape is [m, n].
  • The operator does not support the calculation for three or more dimensions.

Calling Example

For details about the operator calling example, see Strmm.