ADD file via upload

Untitled1.cpp

@@ -0,0 +1,130 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <arm_neon.h>
+
+// 基础向量加法
+void vector_add_base(float *a, float *b, float *c, int n) {
+for (int i = 0; i < n; i++) {
+c[i] = a[i] + b[i];
+}
+}
+
+// NEON优化向量加法
+void vector_add_neon(float *a, float *b, float *c, int n) {
+int i;
+for (i = 0; i <= n - 4; i += 4) {
+float32x4_t va = vld1q_f32(&a[i]);
+float32x4_t vb = vld1q_f32(&b[i]);
+float32x4_t vc = vaddq_f32(va, vb);
+vst1q_f32(&c[i], vc);
+}
+for (; i < n; i++) {
+c[i] = a[i] + b[i];
+}
+}
+
+// 基础矩阵乘法
+void matrix_multiply_base(float *A, float *B, float *C, int N) {
+for (int i = 0; i < N; i++) {
+for (int j = 0; j < N; j++) {
+float sum = 0.0f;
+for (int k = 0; k < N; k++) {
+sum += A[i * N + k] * B[k * N + j];
+}
+C[i * N + j] = sum;
+}
+}
+}
+
+// NEON优化矩阵乘法
+void matrix_multiply_neon(float *A, float *B, float *C, int N) {
+int i, j, k;
+for (i = 0; i < N; i++) {
+for (j = 0; j < N; j++) {
+float32x4_t sum = vdupq_n_f32(0.0f);
+for (k = 0; k <= N - 4; k += 4) {
+float32x4_t va = vld1q_f32(&A[i * N + k]);
+float32x4_t vb = vld1q_f32(&B[k * N + j]);
+sum = vmlaq_f32(sum, va, vb);
+}
+C[i * N + j] = vgetq_lane_f32(sum, 0) + vgetq_lane_f32(sum, 1) +
+vgetq_lane_f32(sum, 2) + vgetq_lane_f32(sum, 3);
+
+// 处理剩余的元素
+for (; k < N; k++) {
+C[i * N + j] += A[i * N + k] * B[k * N + j];
+}
+}
+}
+}
+
+// 初始化矩阵
+void initialize_matrix(float *matrix, int N) {
+for (int i = 0; i < N * N; i++) {
+matrix[i] = (float)(rand() % 100) / 100.0f;
+}
+}
+
+// 计算时间
+double get_time() {
+struct timespec ts;
+clock_gettime(CLOCK_MONOTONIC, &ts);
+return ts.tv_sec + ts.tv_nsec * 1e-9;
+}
+
+int main() {
+int N = 1024; // 矩阵大小
+float *a, *b, *c, *A, *B, *C;
+
+// 分配内存
+a = (float *)aligned_alloc(16, N * sizeof(float));
+b = (float *)aligned_alloc(16, N * sizeof(float));
+c = (float *)aligned_alloc(16, N * sizeof(float));
+A = (float *)aligned_alloc(16, N * N * sizeof(float));
+B = (float *)aligned_alloc(16, N * N * sizeof(float));
+C = (float *)aligned_alloc(16, N * N * sizeof(float));
+
+// 初始化向量和矩阵
+for (int i = 0; i < N; i++) {
+a[i] = (float)(rand() % 100) / 100.0f;
+b[i] = (float)(rand() % 100) / 100.0f;
+}
+initialize_matrix(A, N);
+initialize_matrix(B, N);
+
+// 基础向量加法
+double start_time = get_time();
+vector_add_base(a, b, c, N);
+double end_time = get_time();
+printf("基础向量加法时间: %.6f 秒\n", end_time - start_time);
+
+// NEON优化向量加法
+start_time = get_time();
+vector_add_neon(a, b, c, N);
+end_time = get_time();
+printf("NEON优化向量加法时间: %.6f 秒\n", end_time - start_time);
+
+// 基础矩阵乘法
+start_time = get_time();
+matrix_multiply_base(A, B, C, N);
+end_time = get_time();
+printf("基础矩阵乘法时间: %.6f 秒\n", end_time - start_time);
+
+// NEON优化矩阵乘法
+start_time = get_time();
+matrix_multiply_neon(A, B, C, N);
+end_time = get_time();
+printf("NEON优化矩阵乘法时间: %.6f 秒\n", end_time - start_time);
+
+// 释放内存
+free(a);
+free(b);
+free(c);
+free(A);
+free(B);
+free(C);
+
+return 0;
+}
+