Collaboration diagram for [apu.h] APU Matrix Operations:

Functions
void	APUMatrixMult (uint16_t M, uint16_t N, uint16_t P, void pInputA, void pInputB, void *pResult)
	APU accelerator for matrix multiplication of two matrices C = A*B. More...

void	APUMatrixMultHerm (uint16_t M, void pInputA, void pInputB, void *pResult)
	APU accelerator for matrix multiplication of two Hermitian matrices C = A*B. More...

void	APUMatrixMultSym (uint16_t M, void pInputA, void pInputB, void *pResult)
	APU accelerator for matrix multiplication of two symmetric matrices C = A*B. More...

void	APUMatrixSum (uint16_t M, uint16_t N, void pInputA, void pInputB, void *pResult)
	APU accelerator for matrix addition of two matrices C = A+B. More...

void	APUMatrixScalarMult (uint16_t M, uint16_t N, void pInputA, void pInputB, void *pResult)
	APU accelerator for multiplication of a matrix A and a scalar b. More...

void	APUMatrixScalarSum (uint16_t M, uint16_t N, void pInputA, void pInputB, void *pResult)
	APU accelerator for addition of a matrix A and a scalar b. More...

void	APUMatrixNorm (uint16_t M, uint16_t N, void pInput, void pResult)
	APU accelerator for Frobenius norm computation of a matrix. More...

Detailed Description

Close the Doxygen group.

Function Documentation

§ APUMatrixMult()

void APUMatrixMult	(	uint16_t	M,
		uint16_t	N,
		uint16_t	P,
		void *	pInputA,
		void *	pInputB,
		void *	pResult
	)

APU accelerator for matrix multiplication of two matrices C = A*B.

Calculate matrix multiplication of two matrices.

Defined as:

C[MxP] = A[MxN] * B[NxP]

in which A, B, and C are complex matrices with size [MxP], [MxN], and [NxP] respectively.

Parameters

M	number of rows of matrix A
N	number of columns of matrix A (number of rows of matrix B)
P	number of columns of matrix B
pInputA	a pointer to the base of the first input matrix A, in APU memory
pInputB	a pointer to the base of the second input matrix B, in APU memory
pResult	a pointer to where the result will be placed, in APU memory

Returns: None

References APU_API_MATMATMULT, and APU_GET_DATA_MEM_OFFSET.

§ APUMatrixMultHerm()

void APUMatrixMultHerm	(	uint16_t	M,
		void *	pInputA,
		void *	pInputB,
		void *	pResult
	)

APU accelerator for matrix multiplication of two Hermitian matrices C = A*B.

Calculate matrix multiplication of two Hermitian matrices.

Defined as:

C[MxM] = A[MxM] * B[MxM]

in which A, B, and C are Hermitian matrices with the same size [MxM].

Note: APU stores only upper/right triangular part of Hermitian matrices for memory saving. A Hermitian matrix (MxM) will take M*(M+1)/2 element spaces

Parameters

M	number of rows and columns of matrix A/B/C
pInputA	a pointer to the base of the first input matrix A, in APU memory
pInputB	a pointer to the base of the second input matrix B, in APU memory
pResult	a pointer to where the result will be placed, in APU memory

Returns: None

References APU_API_HERMATRIXMULT, and APU_GET_DATA_MEM_OFFSET.

§ APUMatrixMultSym()

void APUMatrixMultSym	(	uint16_t	M,
		void *	pInputA,
		void *	pInputB,
		void *	pResult
	)

APU accelerator for matrix multiplication of two symmetric matrices C = A*B.

Calculate matrix multiplication of two symmetric matrices.

Defined as:

C[MxM] = A[MxM] * B[MxM]

in which A, B are symmetric matrices with the same size [MxM].

Note: APU stores only upper/right triangular part of symmetric matrices for memory saving. A symmetric matrix (MxM) will take M*(M+1)/2 element spaces

Parameters

M	number of rows and columns of matrix A/B/C
pInputA	a pointer to the base of the first input matrix A, in APU memory
pInputB	a pointer to the base of the second input matrix B, in APU memory
pResult	a pointer to where the result will be placed, in APU memory

Returns: None

References APU_API_SYMMATRIXMULT, and APU_GET_DATA_MEM_OFFSET.

§ APUMatrixSum()

void APUMatrixSum	(	uint16_t	M,
		uint16_t	N,
		void *	pInputA,
		void *	pInputB,
		void *	pResult
	)

APU accelerator for matrix addition of two matrices C = A+B.

Calculate matrix addition of two matrices.

Defined as:

C[MxN] = A[MxN] + B[MxN]

in which, A, B, and C are matrices with the same size [MxN].

Parameters

M	number of rows of matrix A/B/C
N	number of columns of matrix A/B/C
pInputA	a pointer to the base of the first input matrix A, in APU memory
pInputB	a pointer to the base of the second input matrix B, in APU memory
pResult	a pointer to where the result will be placed, in APU memory

Returns: None

References APU_API_MATRIXSUM, and APU_GET_DATA_MEM_OFFSET.

§ APUMatrixScalarMult()

void APUMatrixScalarMult	(	uint16_t	M,
		uint16_t	N,
		void *	pInputA,
		void *	pInputB,
		void *	pResult
	)

APU accelerator for multiplication of a matrix A and a scalar b.

Calculate multiplication of a matrix A and a scalar b.

Defined as:

C[MxN] = b * A[MxN]

in which, A and C are complex matrices with the same size [MxN], b is complex scalar.

Parameters

M	number of rows of matrix A
N	number of columns of matrix A
pInputA	a pointer to the base of the first input matrix A, in APU memory
pInputB	a pointer to the base of the scalar b, in APU memory
pResult	a pointer to where the result will be placed, in APU memory

Returns: None

References APUVectorScalarMult().

§ APUMatrixScalarSum()

void APUMatrixScalarSum	(	uint16_t	M,
		uint16_t	N,
		void *	pInputA,
		void *	pInputB,
		void *	pResult
	)

APU accelerator for addition of a matrix A and a scalar b.

Calculate addition of a matrix A and a scalar b.

Defined as:

C[MxN] = b + A[MxN] (add scalar b to each of A's elements)

in which, A and C are complex matrices with the same size [MxN], b is a complex scalar.

Parameters

M	number of rows of matrix A/C
N	number of columns of matrix A/C
pInputA	a pointer to the base of the first input matrix A, in APU memory
pInputB	a pointer to the base of the scalar b, in APU memory
pResult	a pointer to where the result will be placed, in APU memory

Returns: None

References APUVectorScalarSum().

§ APUMatrixNorm()

void APUMatrixNorm	(	uint16_t	M,
		uint16_t	N,
		void *	pInput,
		void *	pResult
	)

APU accelerator for Frobenius norm computation of a matrix.

Calculates the Frobenius norm of a matrix.

Defined as:

c = ||X|| = sqrt(sum(xij * xij')), i = 1 to M and j = 1 to N

Parameters

M	number of rows of input matrix
N	number of columns of input matrix
pInput	a pointer to the base of the input matrix, in APU memory
pResult	a pointer to output norm value, in APU memory

Returns: None

References APU_API_MATRIXNORM, and APU_GET_DATA_MEM_OFFSET.

Functions

Detailed Description

Function Documentation

§ APUMatrixMult()

§ APUMatrixMultHerm()

§ APUMatrixMultSym()

§ APUMatrixSum()

§ APUMatrixScalarMult()

§ APUMatrixScalarSum()

§ APUMatrixNorm()