HCgemvBatched

Applicable Products

Function Description

API function:

• asdBlasMakeHCgemvBatchedPlan: initializes the operator configuration corresponding to the handle.
• asdBlasHCgemvBatched: computes the product of a batch of complex matrices and vectors.

Formula:

• asdBlasHCgemvBatched formula:

  op indicates that matrix A is conjugately transposed or not transposed.

• Example:

  Input inTensorA[i]:

  [   [ 1+i, 1+2i ],
      [ 1+3i, 1+4i ]  ]

  Input x[i]:

  [ 1+i, 1+i ]

  The input trans is N, indicating that matrix A is not transposed.

  The input m is 2, the input n is 2, the input alpha is 1+0i, and the input beta is 0+0i.

  The input lda is 2.

  The input batchCount is 1.

  After the asdBlasHCgemvBatched operator is called, the output y[i] is as follows:

  [0+4i, 0+8i]

Function Prototype

• AspbStatus asdBlasMakeHCgemvBatchedPlan(asdBlasHandle handle, asdBlasOperation_t trans, const int64_t m)
• AspbStatus asdBlasHCgemvBatched(asdBlasHandle handle, asdBlasOperation_t trans, const int64_t m, const int64_t n,const std::complex<op::fp16_t> &alpha, aclTensor *A, const int64_t lda,aclTensor *x, const int64_t incx, const std::complex<op::fp16_t> &beta,aclTensor *y, const int64_t incy, const int64_t batchCount)

Parameter Description

• asdBlasMakeHCgemvBatchedPlan

  Parameter	Input/Output	Type	Description
  handle	Input	asdBlasHandle	Operator handle.
  trans	Input	asdBlasOperation_t	Whether matrix A should be transposed. ASDBLAS_OP_N //Not transposed ASDBLAS_OP_C //Conjugately transposed
  m	Input	const int64_t	Number of rows in matrix A in a single batch.

• asdBlasHCgemvBatched

  Parameter	Input/Output	Type	Description
  handle	Input	asdBlasHandle	Operator handle.
  trans	Input	asdBlasOperation_t	Whether matrix A should be transposed. ASDBLAS_OP_N //Not transposed ASDBLAS_OP_C //Conjugately transposed
  m	Input	const int64_t	Number of rows in matrix A in a single batch.
  n	Input	const int64_t	Number of columns in matrix A in a single batch.
  alpha	Input	const std::complex<op::fp16_t> &	alpha in the formula is a complex scalar, and is used to multiply the result of the matrix and vector multiplication. In the current version, the value of alpha can only be 1+0i.
  A	Input	aclTensor *	Matrix A in the formula, which is in row-major order and is a tensor on the device. The data type is COMPLEX32, the data format is ND, and the shape is [batchCount, m, n].
  lda	Input	const int64_t	Memory address increment between consecutive elements of A (currently restricted to m).
  x	Input	aclTensor *	Vector x in the formula, which is a tensor on the device. The data type must be COMPLEX32 and the data format must be ND. When the value of trans is not transposed, shape is [batchCount, n]. When the value of trans is conjugately transposed, shape is [batchCount, m].
  incx	Input	const int64_t	Stride of vector x (currently restricted to 1).
  beta	Input	const std::complex<op::fp16_t> &	beta in the formula is a complex scalar, and is used to multiply vector y. In the current version, the value of beta can only be 0+0i.
  y	Input/Output	aclTensor *	Vector y in the formula, which is a tensor on the device. The data type must be COMPLEX64 and the data format must be ND. When the value of trans is not transposed, shape is [batchCount, m]. When the value of trans is conjugately transposed, shape is [batchCount, n].
  incy	Input	const int64_t	Stride of vector y (currently restricted to 1).
  batchCount	Input	const int64_t	Number of batches. The value range is {12 - 314496}.

Return Value Description

For details about the return values, see Return Value.

Constraints

• asdBlasMakeCgemvBatchedPlan: none.
• asdBlasCgemvBatched
  • The supported CANN version is CANN 8.0 or later.
  • The supported input data type is COMPLEX32.
  • The supported output data type is COMPLEX32.

Calling Example

For details about the operator calling example, see HcgemvBatched.