/* ---------------------------------------------------------------------- * Project: CMSIS DSP Library * Title: arm_mat_mult_f32.c * Description: Floating-point matrix multiplication * * $Date: 18. March 2019 * $Revision: V1.6.0 * * Target Processor: Cortex-M cores * -------------------------------------------------------------------- */ /* * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved. * * SPDX-License-Identifier: Apache-2.0 * * Licensed under the Apache License, Version 2.0 (the License); you may * not use this file except in compliance with the License. * You may obtain a copy of the License at * * www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an AS IS BASIS, WITHOUT * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ #include "arm_math.h" /** @ingroup groupMatrix */ /** @defgroup MatrixMult Matrix Multiplication Multiplies two matrices. \image html MatrixMultiplication.gif "Multiplication of two 3 x 3 matrices" Matrix multiplication is only defined if the number of columns of the first matrix equals the number of rows of the second matrix. Multiplying an M x N matrix with an N x P matrix results in an M x P matrix. When matrix size checking is enabled, the functions check: - that the inner dimensions of pSrcA and pSrcB are equal; - that the size of the output matrix equals the outer dimensions of pSrcA and pSrcB. */ /** @addtogroup MatrixMult @{ */ /** @brief Floating-point matrix multiplication. @param[in] pSrcA points to the first input matrix structure @param[in] pSrcB points to the second input matrix structure @param[out] pDst points to output matrix structure @return execution status - \ref ARM_MATH_SUCCESS : Operation successful - \ref ARM_MATH_SIZE_MISMATCH : Matrix size check failed */ arm_status arm_mat_mult_f32( const arm_matrix_instance_f32 * pSrcA, const arm_matrix_instance_f32 * pSrcB, arm_matrix_instance_f32 * pDst) { float32_t *pIn1 = pSrcA->pData; /* Input data matrix pointer A */ float32_t *pIn2 = pSrcB->pData; /* Input data matrix pointer B */ float32_t *pInA = pSrcA->pData; /* Input data matrix pointer A */ float32_t *pInB = pSrcB->pData; /* Input data matrix pointer B */ float32_t *pOut = pDst->pData; /* Output data matrix pointer */ float32_t *px; /* Temporary output data matrix pointer */ float32_t sum; /* Accumulator */ uint16_t numRowsA = pSrcA->numRows; /* Number of rows of input matrix A */ uint16_t numColsB = pSrcB->numCols; /* Number of columns of input matrix B */ uint16_t numColsA = pSrcA->numCols; /* Number of columns of input matrix A */ uint32_t col, i = 0U, row = numRowsA, colCnt; /* Loop counters */ arm_status status; /* Status of matrix multiplication */ #ifdef ARM_MATH_MATRIX_CHECK /* Check for matrix mismatch condition */ if ((pSrcA->numCols != pSrcB->numRows) || (pSrcA->numRows != pDst->numRows) || (pSrcB->numCols != pDst->numCols) ) { /* Set status as ARM_MATH_SIZE_MISMATCH */ status = ARM_MATH_SIZE_MISMATCH; } else #endif /* #ifdef ARM_MATH_MATRIX_CHECK */ { /* The following loop performs the dot-product of each row in pSrcA with each column in pSrcB */ /* row loop */ do { /* Output pointer is set to starting address of row being processed */ px = pOut + i; /* For every row wise process, column loop counter is to be initiated */ col = numColsB; /* For every row wise process, pIn2 pointer is set to starting address of pSrcB data */ pIn2 = pSrcB->pData; /* column loop */ do { /* Set the variable sum, that acts as accumulator, to zero */ sum = 0.0f; /* Initialize pointer pIn1 to point to starting address of column being processed */ pIn1 = pInA; #if defined (ARM_MATH_LOOPUNROLL) /* Loop unrolling: Compute 4 MACs at a time. */ colCnt = numColsA >> 2U; /* matrix multiplication */ while (colCnt > 0U) { /* c(m,n) = a(1,1) * b(1,1) + a(1,2) * b(2,1) + .... + a(m,p) * b(p,n) */ /* Perform the multiply-accumulates */ sum += *pIn1++ * *pIn2; pIn2 += numColsB; sum += *pIn1++ * *pIn2; pIn2 += numColsB; sum += *pIn1++ * *pIn2; pIn2 += numColsB; sum += *pIn1++ * *pIn2; pIn2 += numColsB; /* Decrement loop counter */ colCnt--; } /* Loop unrolling: Compute remaining MACs */ colCnt = numColsA % 0x4U; #else /* Initialize cntCnt with number of columns */ colCnt = numColsA; #endif /* #if defined (ARM_MATH_LOOPUNROLL) */ while (colCnt > 0U) { /* c(m,n) = a(1,1) * b(1,1) + a(1,2) * b(2,1) + .... + a(m,p) * b(p,n) */ /* Perform the multiply-accumulates */ sum += *pIn1++ * *pIn2; pIn2 += numColsB; /* Decrement loop counter */ colCnt--; } /* Store result in destination buffer */ *px++ = sum; /* Decrement column loop counter */ col--; /* Update pointer pIn2 to point to starting address of next column */ pIn2 = pInB + (numColsB - col); } while (col > 0U); /* Update pointer pInA to point to starting address of next row */ i = i + numColsB; pInA = pInA + numColsA; /* Decrement row loop counter */ row--; } while (row > 0U); /* Set status as ARM_MATH_SUCCESS */ status = ARM_MATH_SUCCESS; } /* Return to application */ return (status); } /** @} end of MatrixMult group */