BLASなんもわからん

test_cblas.c

#include <cblas_openblas.h>
#include <stdio.h>
#include <math.h>

int main(void)
{
    double A[3][3] = {
        {1.0, 2.0, 3.0},
        {2.0, 3.0, 4.0},
        {4.0, 5.0, 6.0},
    };
    double B[3][4] = {
        {1.0, 2.0, 3.0, 2.0},
        {2.0, -2.0, 4.0, 0.0},
        {1.0, 5.0, 6.0, -1.0},
    };
    double C[3][4];
    printf("A: {");
    for (size_t i = 0; i < 3 * 3; ++i) {
        printf("%g, ", (&A[0][0])[i]);
    }
    puts("}");
    for (size_t i = 0; i < 3; ++i) {
        for (size_t j = 0; j < 3; ++j) {
            printf("%g, ", A[i][j]); // 3 * i + j
        }
        puts("");
    }
    puts("---");
    printf("sizeof(B[0]) = %zu\n", sizeof(B[0]));
    printf("B: {");
    for (size_t i = 0; i < 3 * 4; ++i) {
        printf("%g, ", (&B[0][0])[i]);
    }
    puts("}");
    for (size_t j = 0; j < 3; ++j) {
        for (size_t k = 0; k < 4; ++k) {
            printf("%g, ", B[j][k]); // 4 * j + k
        }
        puts("");
    }
    puts("---");
    for (size_t i = 0; i < 3; ++i) {
        for (size_t k = 0; k < 4; ++k) {
            double x = 0.0;
            for (size_t j = 0; j < 3; ++j) {
                x += A[i][j] * B[j][k];
            }
            C[i][k] = x;
        }
    }
    printf("C (naive): {");
    for (size_t i = 0; i < 3 * 4; ++i) {
        printf("%g, ", (&C[0][0])[i]);
    }
    puts("}");
    for (size_t i = 0; i < 3; ++i) {
        for (size_t k = 0; k < 4; ++k) {
            printf("%g, ", C[i][k]); // 4 * i + k
        }
        puts("");
    }
    puts("---");
    for (size_t i = 0; i < 3; ++i) {
        for (size_t k = 0; k < 4; ++k) {
            C[i][k] = NAN;
        }
    }
    cblas_dgemm(CblasRowMajor /* or CblasColMajor */, CblasNoTrans, CblasNoTrans, /* m */ 3, /* n */ 4, /* k */ 3, /* alpha */ 1.0, &A[0][0], /* lda */ 3, &B[0][0], /* ldb */ 4, /* beta */ 0.0, &C[0][0], /* ldc */ 4);
    printf("C (blas): {");
    for (size_t i = 0; i < 3 * 4; ++i) {
        printf("%g, ", (&C[0][0])[i]);
    }
    puts("}");
    for (size_t i = 0; i < 3; ++i) {
        for (size_t k = 0; k < 4; ++k) {
            printf("%g, ", C[i][k]); // 4 * i + k
        }
        puts("");
    }
}