Skip to content

Add restrict qualifiers to matrix function calls #1

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ColeStrickler wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add restrict qualifiers to matrix function calls #1

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 29 additions & 8 deletions matrix.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,7 +91,10 @@ double print_checksum(float *C, int dimention)


// a naive matrix multiplication implementation.
void matmult_opt0_naive(float *A, float *B, float *C, int dimension)
void matmult_opt0_naive( const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension)
{
for(int i = 0; i < dimension; i++) {
for(int j = 0; j < dimension; j++) {
Expand All @@ -103,7 +106,10 @@ void matmult_opt0_naive(float *A, float *B, float *C, int dimension)
}

// matrix multiplication with jk order switch
void matmult_opt1_jk(float *A, float *B, float *C, int dimension)
void matmult_opt1_jk(const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension)
{
for(int i = 0; i < dimension; i++) {
for(int k = 0; k < dimension; k++) {
Expand All @@ -116,7 +122,10 @@ void matmult_opt1_jk(float *A, float *B, float *C, int dimension)

// matrix multiplication with jk order switch and tiling
// Handles tail tiles when dimension is not a multiple of block size.
void matmult_opt2_jk_tiling(float *A, float *B, float *C, int dimension)
void matmult_opt2_jk_tiling(const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension)
{
int i,j,k,ii,jj,kk;
int bs = 256; // block size = 256*256*4 = 256KB
Expand Down Expand Up @@ -152,15 +161,18 @@ void transpose_naive(float *src, float *dst, int src_row, int src_col)
}

// matrix multiplicaiton after transposed
void matmult_opt3_transposed(float *A, float *B, float *C, int dimension)
void matmult_opt3_transposed(const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension)
{
int i,j,k;
size_t alloc_size;
if (calc_matrix_bytes(dimension, &alloc_size) != 0) {
fprintf(stderr, "Invalid dimension for allocation\n");
return;
}
float *Bt = (float*)malloc(alloc_size);
float* __restrict__ Bt = (float*)malloc(alloc_size);
if (!Bt) {
fprintf(stderr, "Failed to allocate memory\n");
return;
Expand All @@ -182,7 +194,10 @@ void matmult_opt3_transposed(float *A, float *B, float *C, int dimension)
#ifdef __AVX2__
#include <immintrin.h> // AVX2 Intrinsics
// matrix multiplicaiton transposed with AVX2 SIMD
void matmult_opt4_transposed_simd(float* A, float* B, float* C, int dimension) {
void matmult_opt4_transposed_simd(const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension) {

size_t alloc_size;
if (calc_matrix_bytes(dimension, &alloc_size) != 0) {
Expand Down Expand Up @@ -232,7 +247,10 @@ void matmult_opt4_transposed_simd(float* A, float* B, float* C, int dimension) {
#include <smmintrin.h> // SSE4.2 Intrinsics

// matrix multiplicaiton transposed with SIMD
void matmult_opt4_transposed_simd(float* A, float* B, float* C, int dimension) {
void matmult_opt4_transposed_simd(const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension) {

size_t alloc_size;
if (calc_matrix_bytes(dimension, &alloc_size) != 0) {
Expand Down Expand Up @@ -273,7 +291,10 @@ void matmult_opt4_transposed_simd(float* A, float* B, float* C, int dimension) {
#elif __ARM_NEON
#include <arm_neon.h>
// matrix multiplicaiton transposed with SIMD
void matmult_opt4_transposed_simd(float* A, float* B, float* C, int dimension) {
void matmult_opt4_transposed_simd(const float* __restrict__ A,
const float* __restrict__ B,
float* __restrict__ C,
int dimension) {

size_t alloc_size;
if (calc_matrix_bytes(dimension, &alloc_size) != 0) {
Expand Down