Skip to content

/github/workspace/src/MatrixFunctionsStride/mat_add_stride/kernels/plp_mat_add_stride_i8p_xpulpv2.c

Functions

Name
void plp_mat_add_stride_i8p_xpulpv2(void * args)
Parallel strided matrix addition of 8-bit integer matrices kernel for XPULPV2 extension.

Functions Documentation

function plp_mat_add_stride_i8p_xpulpv2

void plp_mat_add_stride_i8p_xpulpv2(
    void * args
)

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

Parameters:

Return: none

Par: Exploiting SIMD instructions

The 8 bit values are packed four each into 32 bit vectors and then the four dot products are performed on 32 bit vectors, with 32 bit accumulator.

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

Source code

/* =====================================================================
 * Project:      PULP DSP Library
 * Title:        plp_mat_add_stride_i8p_xpulpv2.c
 * Description:  parallel 8-bit integer strided 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_stride_i8p_xpulpv2(void *args) {

    int core_id = hal_core_id();

    plp_mat_add_stride_instance_i8 *a = (plp_mat_add_stride_instance_i8 *)args;

    const int8_t *__restrict__ pSrcA = a->pSrcA;
    const int8_t *__restrict__ pSrcB = a->pSrcB;
    uint32_t M = a->M;
    uint32_t N = a->N;
    uint32_t strideA = a->strideA;
    uint32_t strideB = a->strideB;
    uint32_t strideY = a->strideY;
    uint32_t nPE = a->nPE;
    int8_t *__restrict__ pDst = a->pDst;

//#define BASIC_VERSION // if used don't forget to also use the undefine at end of file
#ifdef BASIC_VERSION

    uint32_t m, n; // loop counters

    for (m = 0; m < M; m++) {
        for (n = 0; n < N; n++) {
            pDst[m * strideY + n] = pSrcA[m * strideA + n] + pSrcB[m * strideB + n];
        }
    }

#else

    uint32_t m, n; // loop counters

    unsigned int n_iter = N >> 2;
    unsigned int n_rem = N & 0x3;

    pSrcA += strideA * core_id;
    pSrcB += strideB * core_id;
    pDst += strideY * core_id;

    unsigned int step_a = strideA * nPE - N;
    unsigned int step_b = strideB * nPE - N;
    unsigned int step_y = strideY * nPE - N;

    for (m = core_id; m < M; m += nPE) {
        for (n = 0; n < n_iter; n++) {
            v4s a = *((v4s *)pSrcA);
            v4s b = *((v4s *)pSrcB);
            *((v4s *)pDst) = __ADD4(a, b);
            pSrcA += 4;
            pSrcB += 4;
            pDst += 4;
        }
        for (n = 0; n < n_rem; n++) {
            *pDst++ = *pSrcA++ + *pSrcB++;
        }
        pSrcA += step_a;
        pSrcB += step_b;
        pDst += step_y;
    }

#endif
    //#undef BASIC_VERSION
}

Updated on 2023-03-01 at 16:16:33 +0000