Skip to content
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.

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

Add support for generic matrix multiplication #25

Open
wants to merge 9 commits into
base: main
Choose a base branch
from
28 changes: 28 additions & 0 deletions include/nn_dot_product_matrix.h
Original file line number Diff line number Diff line change
@@ -0,0 +1,28 @@
#ifndef NN_DOT_PRODUCT_MATRIX_H
#define NN_DOT_PRODUCT_MATRIX_H

#include "nn_error.h"
#include <stddef.h>

// NN_MATRIX_MAX_ROWS defines the maximum number of rows in a matrix.
#ifndef NN_MATRIX_MAX_ROWS
#define NN_MATRIX_MAX_ROWS 3
#endif

// NN_MATRIX_MAX_COLS defines the maximum number of columns in a matrix.
#ifndef NN_MATRIX_MAX_COLS
#define NN_MATRIX_MAX_COLS 3
#endif

// NNDotProductMatrixFunction represents a function that calculates
// the dot product of two matrices.
typedef void (*NNDotProductMatrixFunction)(float result[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS], const float a[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS], const float b[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS]);

// nn_dot_product_matrix calculates the dot product of two square
// matrices.
//
// The dimensions of the input matrices and the resultant matrix are
// implicitly the same.
void nn_dot_product_matrix(float result[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS], const float a[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS], const float b[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS]);

#endif // NN_DOT_PRODUCT_MATRIX_H
25 changes: 25 additions & 0 deletions src/nn_dot_product matrix.c
Original file line number Diff line number Diff line change
@@ -0,0 +1,25 @@
#define NN_DOT_PRODUCT_MATRIX_C
evanmcclure marked this conversation as resolved.
Show resolved Hide resolved
#include "nn_dot_product_matrix.h"
#include "nn_debug.h"
#include <stddef.h>
#include <string.h>

// nn_dot_product_matrix calculates the dot product of two square
// matrices.
void nn_dot_product_matrix(float result[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS], const float a[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS], const float b[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS]) {
NN_DEBUG_PRINT(5, "function %s called\n", __func__);

// Initialize the result matrix.
for (int i = 0; i < NN_MATRIX_MAX_ROWS; i++) {
memset(&result[i], 0, NN_MATRIX_MAX_COLS * sizeof(float));
}

// Multiply two square matrices.
for (int i = 0; i < NN_MATRIX_MAX_ROWS; i++) {
for (int j = 0; j < NN_MATRIX_MAX_COLS; j++) {
for (int k = 0; k < NN_MATRIX_MAX_ROWS; k++) {
Copy link
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should use the size arguments

result[i][j] = result[i][j] + a[i][k] * b[k][j];
}
}
}
}
116 changes: 116 additions & 0 deletions tests/arch/generic/dot_product_matrix/main.c
Original file line number Diff line number Diff line change
@@ -0,0 +1,116 @@
#include "nn_config.h"
#include "nn_debug.h"
#include "nn_dot_product_matrix.h"
#include <assert.h>
#include <math.h>
#include <stdbool.h>
#include <stdio.h>

// N_TEST_CASES defines the number of test cases.
#define N_TEST_CASES 4
// DEFAULT_OUTPUT_TOLERANCE defines the default tolerance for comparing output values.
#define DEFAULT_OUTPUT_TOLERANCE 0.0001f

// TestCase defines a single test case.
typedef struct {
float a[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS];
float b[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS];
float bias;
NNDotProductMatrixFunction dot_product_matrix_func;
float output_tolerance;
float expected_output[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS];
} TestCase;

// run_test_cases runs the test cases.
void run_test_cases(TestCase *test_cases, int n_cases, char *info, NNDotProductMatrixFunction dot_product_matrix_func) {
for (int i = 0; i < n_cases; ++i) {
TestCase tc = test_cases[i];

float output[NN_MATRIX_MAX_ROWS][NN_MATRIX_MAX_COLS];

NN_DEBUG_PRINT(5, "A:\n");
for (int i = 0; i < NN_MATRIX_MAX_ROWS; i++) {
for (int j = 0; j < NN_MATRIX_MAX_COLS; j++) {
NN_DEBUG_PRINT(5, " %f", tc.a[i][j]);
}
NN_DEBUG_PRINT(5, "\n");
}

NN_DEBUG_PRINT(5, "B:\n");
for (int i = 0; i < NN_MATRIX_MAX_ROWS; i++) {
for (int j = 0; j < NN_MATRIX_MAX_COLS; j++) {
NN_DEBUG_PRINT(5, " %f", tc.b[i][j]);
}
NN_DEBUG_PRINT(5, "\n");
}

dot_product_matrix_func(output, tc.a, tc.b);

NN_DEBUG_PRINT(5, "C:\n");
for (int i = 0; i < NN_MATRIX_MAX_ROWS; i++) {
for (int j = 0; j < NN_MATRIX_MAX_COLS; j++) {
NN_DEBUG_PRINT(5, " %f", tc.expected_output[i][j]);
}
NN_DEBUG_PRINT(5, "\n");
}

for (int m = 0; m < NN_MATRIX_MAX_ROWS; m++) {
for (int n = 0; n < NN_MATRIX_MAX_COLS; n++) {
assert(isnan(output[m][n]) == false);
assert(fabs(output[m][n] - tc.expected_output[m][n]) < tc.output_tolerance);
}
}
printf("passed: %s case=%d info=%s\n", __func__, i + 1, info);
}
}

int main() {
// nn_set_debug_level(10);

TestCase test_cases[N_TEST_CASES] = {
{
.a = {{0}},
.b = {{0}},
.output_tolerance = DEFAULT_OUTPUT_TOLERANCE,
.expected_output = {{0}},
},

{
.a = {{ 3, 0},
{-1, 2},
{ 1, 1}},
.b = {{4, -1},
{0, 2}},
.output_tolerance = DEFAULT_OUTPUT_TOLERANCE,
.expected_output = {{ 12, -3},
{-4, 5},
{ 4, 1}},
},

{
.a = {{ 1, 5, 2},
{-1, 0, 1},
{ 3, 2, 4}},
.b = {{ 6, 1, 3},
{-1, 1, 2},
{ 4, 1, 3}},
.output_tolerance = DEFAULT_OUTPUT_TOLERANCE,
.expected_output = {{ 9, 8, 19},
{-2, 0, 0},
{32, 9, 25}},
},

{
.a = {{1, 2},
{3, 4}},
.b = {{5},
{6}},
.output_tolerance = DEFAULT_OUTPUT_TOLERANCE,
.expected_output = {{17},
{39}},
},

};
run_test_cases(test_cases, N_TEST_CASES, "nn_dot_product_matrix", nn_dot_product_matrix);
return 0;
}