/github/workspace/src/MatrixFunctions/mat_mult/kernels/plp_mat_mult_i8p_xpulpv2.c
Functions
Name | |
---|---|
void | plp_mat_mult_i8p_xpulpv2(void * args) Parallel matrix multiplication of 8-bit integer matrices kernel for XPULPV2 extension. |
Attributes
Name | |
---|---|
HAL_CL_L1 v4s | mask0 Parallel matrix multiplication of 8-bit integer matrices kernel for XPULPV2 extension. |
HAL_CL_L1 v4s | mask1 |
HAL_CL_L1 v4s | mask2 |
HAL_CL_L1 v4s | mask3 |
Functions Documentation
function plp_mat_mult_i8p_xpulpv2
void plp_mat_mult_i8p_xpulpv2(
void * args
)
Parallel matrix multiplication of 8-bit integer matrices kernel for XPULPV2 extension.
Parameters:
- args pointer to plp_mat_mult_instance_i8 struct initialized by plp_mat_mult_i8_parallel
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.
Attributes Documentation
variable mask0
static HAL_CL_L1 v4s mask0 = { 0, 1, 4, 5 };
Parallel matrix multiplication of 8-bit integer matrices kernel for XPULPV2 extension.
Parameters:
- args pointer to plp_mat_mult_instance_i8 struct initialized by plp_mat_mult_i8_parallel
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.
variable mask1
static HAL_CL_L1 v4s mask1 = { 2, 3, 6, 7 };
variable mask2
static HAL_CL_L1 v4s mask2 = { 0, 2, 4, 6 };
variable mask3
static HAL_CL_L1 v4s mask3 = { 1, 3, 5, 7 };
Source code
/* =====================================================================
* Project: PULP DSP Library
* Title: plp_mat_mult_i8p_xpulpv2.c
* Description: parallel 8-bit integer matrix multiplication for XPULPV2
*
* $Date: July 2022
* $Revision: V1
*
* Target Processor: PULP cores
* ===================================================================== */
/*
* Copyright (C) 2019 ETH Zurich and University of Bologna.
*
* Author: Emmet Murphy, 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"
// define BASIC_VERSION // if used don't forget to also use the undefine at end of file
#ifdef BASIC_VERSION
void plp_mat_mult_i8p_xpulpv2(void *args) {
plp_mat_mult_instance_i8 *arguments = (plp_mat_mult_instance_i8 *)args;
const int8_t *__restrict__ pSrcA = arguments->pSrcA;
const int8_t *__restrict__ pSrcB = arguments->pSrcB;
uint32_t M = arguments->M;
uint32_t N = arguments->N;
uint32_t O = arguments->O;
uint32_t nPE = arguments->nPE;
int32_t *__restrict__ pDstC = arguments->pDstC;
uint32_t i; // loop counter
uint32_t j; // loop counter
uint32_t k; // loop counter
for (i = 0; i < M; i++) {
for (k = 0; k < O; k++) {
int32_t sum = 0;
for (j = 0; j < N; j++) {
sum = sum + pSrcA[i * N + j] * pSrcB[j * O + k];
}
pDstC[i * O + k] = sum;
}
}
hal_team_barrier();
}
#else
HAL_CL_L1 static v4s mask0 = { 0, 1, 4, 5 };
HAL_CL_L1 static v4s mask1 = { 2, 3, 6, 7 };
HAL_CL_L1 static v4s mask2 = { 0, 2, 4, 6 };
HAL_CL_L1 static v4s mask3 = { 1, 3, 5, 7 };
void plp_mat_mult_i8p_xpulpv2(void *args) {
plp_mat_mult_instance_i8 *arguments = (plp_mat_mult_instance_i8 *)args;
const int8_t *__restrict__ pSrcA = arguments->pSrcA;
const int8_t *__restrict__ pSrcB = arguments->pSrcB;
uint32_t M = arguments->M;
uint32_t N = arguments->N;
uint32_t O = arguments->O;
uint32_t nPE = arguments->nPE;
int32_t *__restrict__ pDstC = arguments->pDstC;
uint32_t i = 0; // loop counter for M
uint32_t j = 0; // loop counter for N
uint32_t k = 0; // loop counter for O
uint32_t core_id = hal_core_id();
for (k = core_id; k < O / 4; k += nPE) {
// shuffled data not dependent on i
// preshuffle and store
v4s bVecs[(N / 4) * 4];
for (j = 0; j < N / 4; j++) {
v4s temp0 = *((v4s *)&(pSrcB[(j * 4) * O + (k * 4)]));
v4s temp1 = *((v4s *)&(pSrcB[(j * 4 + 1) * O + (k * 4)]));
v4s temp2 = *((v4s *)&(pSrcB[(j * 4 + 2) * O + (k * 4)]));
v4s temp3 = *((v4s *)&(pSrcB[(j * 4 + 3) * O + (k * 4)]));
v4s temp4 = __builtin_shuffle(temp0, temp1, mask0); // 0,1,4,5
v4s temp5 = __builtin_shuffle(temp2, temp3, mask0); // 8,9,12,13
v4s temp6 = __builtin_shuffle(temp0, temp1, mask1); // 2,3,6,7
v4s temp7 = __builtin_shuffle(temp2, temp3, mask1); // 3,7,11,15
bVecs[(j * 4)] = __builtin_shuffle(temp4, temp5, mask2);
bVecs[(j * 4) + 1] = __builtin_shuffle(temp4, temp5, mask3);
bVecs[(j * 4) + 2] = __builtin_shuffle(temp6, temp7, mask2);
bVecs[(j * 4) + 3] = __builtin_shuffle(temp6, temp7, mask3);
}
for (i = 0; i < M / 2; i++) {
int32_t sum00 = 0;
int32_t sum01 = 0;
int32_t sum02 = 0;
int32_t sum03 = 0;
int32_t sum10 = 0;
int32_t sum11 = 0;
int32_t sum12 = 0;
int32_t sum13 = 0;
for (j = 0; j < N / 4; j++) {
v4s aVec0 = *((v4s *)&(pSrcA[(i * 2) * N + (j * 4)]));
v4s aVec1 = *((v4s *)&(pSrcA[(i * 2 + 1) * N + (j * 4)]));
v4s bVec0 = bVecs[(j * 4)];
v4s bVec1 = bVecs[(j * 4) + 1];
v4s bVec2 = bVecs[(j * 4) + 2];
v4s bVec3 = bVecs[(j * 4) + 3];
sum00 = __SUMDOTP4(aVec0, bVec0, sum00);
sum01 = __SUMDOTP4(aVec0, bVec1, sum01);
sum02 = __SUMDOTP4(aVec0, bVec2, sum02);
sum03 = __SUMDOTP4(aVec0, bVec3, sum03);
sum10 = __SUMDOTP4(aVec1, bVec0, sum10);
sum11 = __SUMDOTP4(aVec1, bVec1, sum11);
sum12 = __SUMDOTP4(aVec1, bVec2, sum12);
sum13 = __SUMDOTP4(aVec1, bVec3, sum13);
}
for (j = j * 4; j < N; j++) {
int32_t aVal0 = pSrcA[(i * 2) * N + j];
int32_t aVal1 = pSrcA[(i * 2 + 1) * N + j];
int32_t bVal0 = pSrcB[j * O + (k * 4)];
int32_t bVal1 = pSrcB[j * O + (k * 4) + 1];
int32_t bVal2 = pSrcB[j * O + (k * 4) + 2];
int32_t bVal3 = pSrcB[j * O + (k * 4) + 3];
sum00 += aVal0 * bVal0;
sum01 += aVal0 * bVal1;
sum02 += aVal0 * bVal2;
sum03 += aVal0 * bVal3;
sum10 += aVal1 * bVal0;
sum11 += aVal1 * bVal1;
sum12 += aVal1 * bVal2;
sum13 += aVal1 * bVal3;
}
pDstC[(i * 2) * O + (k * 4)] = sum00;
pDstC[(i * 2) * O + (k * 4 + 1)] = sum01;
pDstC[(i * 2) * O + (k * 4 + 2)] = sum02;
pDstC[(i * 2) * O + (k * 4 + 3)] = sum03;
pDstC[(i * 2 + 1) * O + (k * 4)] = sum10;
pDstC[(i * 2 + 1) * O + (k * 4 + 1)] = sum11;
pDstC[(i * 2 + 1) * O + (k * 4 + 2)] = sum12;
pDstC[(i * 2 + 1) * O + (k * 4 + 3)] = sum13;
}
for (i = i * 2; i < M; i++) {
int32_t sum00 = 0;
int32_t sum01 = 0;
int32_t sum02 = 0;
int32_t sum03 = 0;
for (j = 0; j < N / 4; j++) {
v4s aVec0 = *((v4s *)&(pSrcA[i * N + (j * 4)]));
v4s bVec0 = bVecs[(j * 4)];
v4s bVec1 = bVecs[(j * 4) + 1];
v4s bVec2 = bVecs[(j * 4) + 2];
v4s bVec3 = bVecs[(j * 4) + 3];
sum00 = __SUMDOTP4(aVec0, bVec0, sum00);
sum01 = __SUMDOTP4(aVec0, bVec1, sum01);
sum02 = __SUMDOTP4(aVec0, bVec2, sum02);
sum03 = __SUMDOTP4(aVec0, bVec3, sum03);
}
for (j = j * 4; j < N; j++) {
int32_t aVal0 = pSrcA[i * N + j];
int32_t bVal0 = pSrcB[j * O + (k * 4)];
int32_t bVal1 = pSrcB[j * O + (k * 4) + 1];
int32_t bVal2 = pSrcB[j * O + (k * 4) + 2];
int32_t bVal3 = pSrcB[j * O + (k * 4) + 3];
sum00 += aVal0 * bVal0;
sum01 += aVal0 * bVal1;
sum02 += aVal0 * bVal2;
sum03 += aVal0 * bVal3;
}
pDstC[i * O + (k * 4)] = sum00;
pDstC[i * O + (k * 4 + 1)] = sum01;
pDstC[i * O + (k * 4 + 2)] = sum02;
pDstC[i * O + (k * 4 + 3)] = sum03;
}
}
for (k = k * 4; k < O; k++) {
for (i = 0; i < M / 4; i++) {
int32_t sum0 = 0;
int32_t sum1 = 0;
int32_t sum2 = 0;
int32_t sum3 = 0;
for (j = 0; j < N; j++) {
int32_t aVal0 = pSrcA[(i * 4) * N + j];
int32_t aVal1 = pSrcA[(i * 4 + 1) * N + j];
int32_t aVal2 = pSrcA[(i * 4 + 2) * N + j];
int32_t aVal3 = pSrcA[(i * 4 + 3) * N + j];
int32_t bVal0 = pSrcB[j * O + k];
sum0 += aVal0 * bVal0;
sum1 += aVal1 * bVal0;
sum2 += aVal2 * bVal0;
sum3 += aVal3 * bVal0;
}
pDstC[(i * 4) * O + k] = sum0;
pDstC[(i * 4 + 1) * O + k] = sum1;
pDstC[(i * 4 + 2) * O + k] = sum2;
pDstC[(i * 4 + 3) * O + k] = sum3;
}
for (i = i * 4; i < M; i++) {
int32_t sum0 = 0;
for (j = 0; j < N; j++) {
int32_t aVal0 = pSrcA[i * N + j];
int32_t bVal0 = pSrcB[j * O + k];
sum0 += aVal0 * bVal0;
}
pDstC[i * O + k] = sum0;
}
}
hal_team_barrier();
}
#endif
// undefine BASIC_VERSION
Updated on 2023-03-01 at 16:16:33 +0000