Skip to content

/github/workspace/src/MatrixFunctions/mat_add/kernels/plp_mat_add_i8s_xpulpv2.c

Functions

Name
void plp_mat_add_i8s_xpulpv2(const int8_t restrict pSrcA, const int8_t restrict pSrcB, uint32_t M, uint32_t N, int8_t *restrict pDst)
matrix addition of 8-bit integer matrices kernel for XPULPV2 extension.

Functions Documentation

function plp_mat_add_i8s_xpulpv2

void plp_mat_add_i8s_xpulpv2(
    const int8_t *__restrict__ pSrcA,
    const int8_t *__restrict__ pSrcB,
    uint32_t M,
    uint32_t N,
    int8_t *__restrict__ pDst
)

matrix addition of 8-bit integer matrices kernel for XPULPV2 extension.

Parameters:

  • pSrcA Points to the first input matrix
  • pSrcB Points to the second input matrix
  • M Height of all matrices
  • N Width of all matrices
  • pDst Points to the output matrix

Return: none

matrix addition of a 8-bit integer matrices for XPULPV2 extension.

Source code

/* =====================================================================
 * Project:      PULP DSP Library
 * Title:        plp_mat_add_i8s_xpulpv2.c
 * Description:  8-bit integer matrix addition for XPULPV2
 *
 * $Date:        1. July 2020
 * $Revision:    V0
 *
 * Target Processor: PULP cores
 * ===================================================================== */
/*
 * Copyright (C) 2020 ETH Zurich and University of Bologna.
 *
 * Author: Tibor Schneider, ETH Zurich
 *
 * 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 "plp_math.h"

void plp_mat_add_i8s_xpulpv2(const int8_t *__restrict__ pSrcA,
                             const int8_t *__restrict__ pSrcB,
                             uint32_t M,
                             uint32_t N,
                             int8_t *__restrict__ pDst) {

    uint32_t i; // loop counters
    uint32_t total = M*N; // we can see it as a 1D operation
#if defined(PLP_MATH_LOOPUNROLL)
    // loop over the matrix - the shift by one is for the loop unrolling
    for (i = 0; i < total>>3; i++) {
        *((v4s*)(pDst + 8*i    )) = __ADD4(*((v4s*)(pSrcA + 8*i    )), *((v4s*)(pSrcB + 8*i    )));
        *((v4s*)(pDst + 8*i + 4)) = __ADD4(*((v4s*)(pSrcA + 8*i + 4)), *((v4s*)(pSrcB + 8*i + 4)));
    }
    // to save the branch we just always compute the possibly remaining elements
    *((v4s*)(pDst + total - 4)) = __ADD4(*((v4s*)(pSrcA + total - 4)), *((v4s*)(pSrcB + total - 4)));
    *((v4s*)(pDst + total - 8)) = __ADD4(*((v4s*)(pSrcA + total - 8)), *((v4s*)(pSrcB + total - 8)));
#else // No PLP_MATH_LOOPUNROLL
    for (i = 0; i < total; i++) {
            pDst[i] = pSrcA[i] + pSrcB[i];
    }

#endif // PLP_MATH_LOOPUNROLL
}

Updated on 2023-03-01 at 16:16:32 +0000