#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <arm_neon.h>
#define SIZE 1024 
#define BLOCK_SIZE 4 
void matmul_optimized(float** A, float** B, float** C, int n) {
    for (int i = 0; i < n; i += BLOCK_SIZE) {
        for (int j = 0; j < n; j += BLOCK_SIZE) {
            float32x4_t vecC[BLOCK_SIZE][BLOCK_SIZE] = {0}; 
            for (int k = 0; k < n; k += BLOCK_SIZE) {
                for (int ii = 0; ii < BLOCK_SIZE && i + ii < n; ii++) {
                    for (int jj = 0; jj < BLOCK_SIZE && k + jj < n; jj++) {
                        float32x4_t vecA = vld1q_f32(&A[i + ii][k]);
                        float32x4_t vecB = vld1q_f32(&B[k + jj][j]);
                        for (int l = 0; l < BLOCK_SIZE && j + l < n; l++) {
                            vecC[ii][l] = vmlaq_f32(vecC[ii][l], vecA, vecB);
                            vecB = vld1q_f32(&B[k + jj][j + BLOCK_SIZE + l]); 
                        }
                    }
                }
            }
            for (int ii = 0; ii < BLOCK_SIZE && i + ii < n; ii++) {
                for (int jj = 0; jj < BLOCK_SIZE && j + jj < n; jj++) {
                    vst1q_f32(&C[i + ii][j + jj], vecC[ii][jj]);
                }
            }
        }
    }
}
void sparse_to_dense(SparseMatrix* sparse, float denseMatrix[ROWS][COLS], int rows, int cols) {
    memset(denseMatrix, 0, sizeof(float) * rows * cols); 
    for (int i = 0; i < sparse->num_elements; i++) {
        int row = sparse->elements[i].row;
        int col = sparse->elements[i].col;
        float value = sparse->elements[i].value;
        denseMatrix[row][col] = value;
    }
}