ADD file via upload

Yang/1.c

@@ -0,0 +1,48 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+#define SIZE 10000000
+
+void vectorAdd(float* A, float* B, float* C, int size) {
+    int i;
+    for (i = 0; i < size; ++i) {
+        C[i] = A[i] + B[i];
+    }
+}
+
+int main() {
+    float* A = (float*)malloc(SIZE * sizeof(float));
+    float* B = (float*)malloc(SIZE * sizeof(float));
+    float* C = (float*)malloc(SIZE * sizeof(float));
+    if (A == NULL || B == NULL || C == NULL) {
+        return 1; 
+    }
+
+    // float A[SIZE], B[SIZE], C[SIZE];
+    srand((unsigned)time(NULL));
+    int i;
+    for (i = 0; i < SIZE; ++i) {
+        A[i] = rand() % 100; 
+        B[i] = rand() % 100;
+    }
+
+
+    clock_t start_time = clock();
+
+    vectorAdd(A, B, C, SIZE);
+
+    clock_t end_time = clock();
+    double time = (double)(end_time - start_time) / CLOCKS_PER_SEC;
+
+    printf("%f\n", time);
+
+    // for (i = 0; i < SIZE; ++i) { 
+    //     printf("A[%d] + B[%d] = C[%d] -> %f + %f = %f\n", i, i, i, A[i], B[i], C[i]);
+    // }
+
+    free(A);
+    free(B);
+    free(C);
+    return 0;
+}

Yang/2.c

@@ -0,0 +1,59 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <arm_neon.h>
+
+
+#define SIZE 10000000
+
+void vectorAdd(float* A, float* B, float* C, int size) {
+    int i;
+    for (i = 0; i < size; ++i) {
+        C[i] = A[i] + B[i];
+    }
+}
+
+void vectorAddNEON(float* A, float* B, float* C, int size) {
+    int i;
+    for (i = 0; i <= size - 4; i += 4) { 
+        float32x4_t vecA = vld1q_f32(&A[i]);
+        float32x4_t vecB = vld1q_f32(&B[i]);
+
+        float32x4_t vecC = vaddq_f32(vecA, vecB);
+        vst1q_f32(&C[i], vecC);
+    }
+    for (; i < size; ++i) {
+        C[i] = A[i] + B[i];
+    }
+}
+
+int main() {
+    float* A = (float*)malloc(SIZE * sizeof(float));
+    float* B = (float*)malloc(SIZE * sizeof(float));
+    float* C = (float*)malloc(SIZE * sizeof(float));
+    if (A == NULL || B == NULL || C == NULL) {
+        return 1; 
+    }
+
+    srand((unsigned)time(NULL));
+        for (int i = 0; i < SIZE; ++i) {
+            A[i] = (float)(rand() % 100);
+            B[i] = (float)(rand() % 100);
+        }
+
+    clock_t start_time = clock();
+    vectorAdd(A, B, C, SIZE);
+    clock_t end_time = clock();
+    double time1 = (double)(end_time - start_time) / CLOCKS_PER_SEC;
+    printf("tradition: %f", time1);
+
+    start_time = clock();
+    vectorAddNEON(A, B, C, SIZE);
+    end_time = clock();
+    time1 = (double)(end_time - start_time) / CLOCKS_PER_SEC; 
+    printf("NEON %f",time1);
+
+    free(A);
+    free(B);
+    free(C);   
+}

Yang/3.c

@@ -0,0 +1,52 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+#define SIZE 1024
+
+void matmul(float** A, float** B, float** C, int n) {
+    int i, j, k;
+    for (i = 0; i < n; ++i) {
+        for (j = 0; j < n; ++j) {
+            C[i][j] = 0;
+            for (k = 0; k < n; ++k) {
+                C[i][j] += A[i][k] * B[k][j];
+            }
+        }
+    }
+}
+
+int main() {
+    float** A = (float**)malloc(SIZE * sizeof(float*));
+    float** B = (float**)malloc(SIZE * sizeof(float*));
+    float** C = (float**)malloc(SIZE * sizeof(float*));
+    int i, j;
+    for (i = 0; i < SIZE; ++i) {
+        A[i] = (float*)malloc(SIZE * sizeof(float));
+        B[i] = (float*)malloc(SIZE * sizeof(float));
+        C[i] = (float*)malloc(SIZE * sizeof(float));
+    }
+
+    srand(time(0));
+    for (i = 0; i < SIZE; ++i) {
+        for (j = 0; j < SIZE; ++j) {
+            A[i][j] = rand() % 100; 
+            B[i][j] = rand() % 100; 
+        }
+    }
+    clock_t start = clock();
+    matmul(A, B, C, SIZE);
+    clock_t end = clock();
+    double time = (double)(end - start) / CLOCKS_PER_SEC;
+    printf("%f\n", time);
+
+    for (i = 0; i < SIZE; ++i) {
+        free(A[i]);
+        free(B[i]);
+        free(C[i]);
+    }
+    free(A);
+    free(B);
+    free(C);
+    return 0;
+}

Yang/4.c

@@ -0,0 +1,75 @@
+#include <arm_neon.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+#define SIZE 1024
+
+void matmul(float** A, float** B, float** C, int n) {
+    int i, j, k;
+    for (i = 0; i < n; ++i) {
+        for (j = 0; j < n; ++j) {
+            C[i][j] = 0;
+            for (k = 0; k < n; ++k) {
+                C[i][j] += A[i][k] * B[k][j];
+            }
+        }
+    }
+}
+
+void matmulNEON(float** A, float** B, float** C, int n) {
+    for (int i = 0; i < n; i++) {
+        for (int j = 0; j < n; j++) {
+            float32x4_t vecC = vmovq_n_f32(0); 
+            for (int k = 0; k < n; k += 4) {  
+                float32x4_t vecA = vld1q_f32(&A[i][k]);
+                float32x4_t vecB = vld1q_f32(&B[k][j]);
+                vecC = vmlaq_f32(vecC, vecA, vecB);  
+            }
+            C[i][j] = vgetq_lane_f32(vecC, 0) + vgetq_lane_f32(vecC, 1) +
+                       vgetq_lane_f32(vecC, 2) + vgetq_lane_f32(vecC, 3);
+        }
+    }
+}
+
+int main() {
+    float** A = (float**)malloc(SIZE * sizeof(float*));
+    float** B = (float**)malloc(SIZE * sizeof(float*));
+    float** C = (float**)malloc(SIZE * sizeof(float*));
+    int i, j;
+    for (i = 0; i < SIZE; ++i) {
+        A[i] = (float*)malloc(SIZE * sizeof(float));
+        B[i] = (float*)malloc(SIZE * sizeof(float));
+        C[i] = (float*)malloc(SIZE * sizeof(float));
+    }
+
+    srand(time(0));
+    for (i = 0; i < SIZE; ++i) {
+        for (j = 0; j < SIZE; ++j) {
+            A[i][j] = rand() % 100; 
+            B[i][j] = rand() % 100; 
+        }
+    }
+    clock_t start1 = clock();
+    matmul(A, B, C, SIZE);
+    clock_t end1 = clock();
+    double time1 = (double)(end1 - start1) / CLOCKS_PER_SEC;
+    printf("tradition %f\n", time1);
+
+    clock_t start2 = clock();
+    matmulNEON(A, B, C, SIZE);
+    clock_t end2 = clock();
+    double time2 = (double)(end2 - start2) / CLOCKS_PER_SEC;
+    printf("NEON %f\n", time2);
+
+    for (i = 0; i < SIZE; i++) {
+        free(A[i]);
+        free(B[i]);
+        free(C[i]);
+    }
+    free(A);
+    free(B);
+    free(C);
+
+    return 0;
+}

Yang/5(generate).c

@@ -0,0 +1,116 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+typedef struct {
+    int* values;      
+    int* rowIndex;    
+    int* colIndex;    
+    int nonZeroCount; 
+} SparseMatrix;
+
+void sparseMatmul(SparseMatrix* A, SparseMatrix* B, SparseMatrix* C) {
+    int currentIndex = 0;
+    int i, j;
+    for (i = 0; i < A->nonZeroCount; i++)
+    {
+        int rowA = A->rowIndex[i];
+        int colA = A->colIndex[i];
+        float valueA = A->values[i];
+        for (j = 0; j < A->nonZeroCount; j++)
+        {
+            int rowB = B->rowIndex[j];
+            int colB = B->colIndex[j];
+            float valueB = B->values[j];
+            if (colA == rowB)
+            {
+                float product = valueA * valueB;
+                int found = 0;
+                int k;
+                for (k = 0; k < currentIndex; k++)
+                {
+                    if (C->rowIndex[k] == rowA && C->colIndex[k] == colB){
+                        C->values[k] += product;
+                        found = 1;
+                        break;
+                    }
+                }
+                if (!found)
+                {
+                    C->values[currentIndex] = product;
+                    C->rowIndex[currentIndex] = rowA;
+                    C->colIndex[currentIndex] = colB;
+                    currentIndex++;
+                }
+            }
+        }
+    }
+    C->nonZeroCount = currentIndex;
+}
+
+void generate(SparseMatrix* matrix, int rows, int cols, int nonZeroCount){
+    matrix->values = (int*)malloc(sizeof(int) * nonZeroCount);
+    matrix->rowIndex = (int*)malloc(sizeof(int) * nonZeroCount);
+    matrix->colIndex = (int*)malloc(sizeof(int) * nonZeroCount);   
+    matrix->nonZeroCount = nonZeroCount;
+    int i;
+    for (i = 0; i < nonZeroCount; i++)
+    {
+        matrix->rowIndex[i] = rand() % rows;
+        matrix->colIndex[i] = rand() % cols;
+        matrix->values[i] = rand() %100;
+    }
+}
+
+void free_matrix(SparseMatrix* matrix) {
+    free(matrix->values);
+    free(matrix->rowIndex);
+    free(matrix->colIndex);
+}
+
+int main() {
+    srand(time(NULL));
+
+    // int i;
+    int rowsA = 1000;
+    int rowsB = 2000;
+    int colsB = 1000;
+    int nonZeroCountA = 10000;
+    int nonZeroCountB = 10000;
+
+    SparseMatrix A, B;
+    generate(&A, rowsA, rowsB, nonZeroCountA);
+    generate(&B, rowsB, colsB, nonZeroCountB);
+
+    // for (i = 0; i < A.nonZeroCount; ++i) {
+    //     printf("A[%d][%d] = %d\n", A.rowIndex[i], A.colIndex[i], A.values[i]);
+    // }
+
+    // for (i = 0; i < B.nonZeroCount; ++i) {
+    //     printf("B[%d][%d] = %d\n", B.rowIndex[i], B.colIndex[i], B.values[i]);
+    // }
+    
+
+    SparseMatrix C;
+    C.values = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.rowIndex = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.colIndex = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.nonZeroCount = 0;
+
+    clock_t start_time = clock();
+
+    sparseMatmul(&A, &B, &C);
+
+    clock_t end_time = clock();
+    double time = (double)(end_time - start_time) / CLOCKS_PER_SEC;
+    printf("%f\n", time);
+    
+    // for (i = 0; i < C.nonZeroCount; ++i) {
+    //     printf("C[%d][%d] = %d\n", C.rowIndex[i], C.colIndex[i], C.values[i]);
+    // }
+
+    free_matrix(&A);
+    free_matrix(&B);
+    free_matrix(&C);
+    return 0;
+}

Yang/5.c

@@ -0,0 +1,96 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+typedef struct {
+    int* values;      
+    int* rowIndex;    
+    int* colIndex;    
+    int nonZeroCount; 
+} SparseMatrix;
+
+void sparseMatmul(SparseMatrix* A, SparseMatrix* B, SparseMatrix* C) {
+    int currentIndex = 0;
+    int i, j;
+    for (i = 0; i < A->nonZeroCount; i++)
+    {
+        int rowA = A->rowIndex[i];
+        int colA = A->colIndex[i];
+        float valueA = A->values[i];
+        for (j = 0; j < A->nonZeroCount; j++)
+        {
+            int rowB = B->rowIndex[j];
+            int colB = B->colIndex[j];
+            float valueB = B->values[j];
+            if (colA == rowB)
+            {
+                float product = valueA * valueB;
+                int found = 0;
+                int k;
+                for (k = 0; k < currentIndex; k++)
+                {
+                    if (C->rowIndex[k] == rowA && C->colIndex[k] == colB){
+                        C->values[k] += product;
+                        found = 1;
+                        break;
+                    }
+                }
+                if (!found)
+                {
+                    C->values[currentIndex] = product;
+                    C->rowIndex[currentIndex] = rowA;
+                    C->colIndex[currentIndex] = colB;
+                    currentIndex++;
+                }
+            }
+        }
+    }
+    C->nonZeroCount = currentIndex;
+}
+
+void free_matrix(SparseMatrix* matrix) {
+    free(matrix->values);
+    free(matrix->rowIndex);
+    free(matrix->colIndex);
+}
+
+int main() {
+    SparseMatrix A = {
+        .values = (int[]){1, 2, 3, 4, 5},
+        .rowIndex = (int[]){0, 0, 1, 2, 2},
+        .colIndex = (int[]){0, 2, 1, 0, 2},
+        .nonZeroCount = 5
+    };
+    
+    SparseMatrix B = {
+        .values = (int[]){6, 8, 7, 9},
+        .rowIndex = (int[]){0, 2, 1, 2},
+        .colIndex = (int[]){0, 0, 1, 2},
+        .nonZeroCount = 4
+    };
+
+    SparseMatrix C;
+    C.values = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.rowIndex = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.colIndex = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.nonZeroCount = 0;
+
+    clock_t start_time = clock();
+
+    sparseMatmul(&A, &B, &C);
+
+    clock_t end_time = clock();
+    double time = (double)(end_time - start_time) / CLOCKS_PER_SEC;
+    printf("%f\n", time);
+    
+    int i;
+    for (i = 0; i < C.nonZeroCount; ++i) {
+        printf("C[%d][%d] = %d\n", C.rowIndex[i], C.colIndex[i], C.values[i]);
+    }
+
+    free_matrix(&A);
+    free_matrix(&B);
+    free_matrix(&C);
+
+    return 0;
+}

Yang/6.c

@@ -0,0 +1,155 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <arm_neon.h>
+
+typedef struct {
+    int* values;      
+    int* rowIndex;    
+    int* colIndex;    
+    int nonZeroCount; 
+} SparseMatrix;
+
+void sparseMatmul(SparseMatrix* A, SparseMatrix* B, SparseMatrix* C) {
+    int currentIndex = 0;
+    int i, j;
+    for (i = 0; i < A->nonZeroCount; i++)
+    {
+        int rowA = A->rowIndex[i];
+        int colA = A->colIndex[i];
+        float valueA = A->values[i];
+        for (j = 0; j < A->nonZeroCount; j++)
+        {
+            int rowB = B->rowIndex[j];
+            int colB = B->colIndex[j];
+            float valueB = B->values[j];
+            if (colA == rowB)
+            {
+                float product = valueA * valueB;
+                int found = 0;
+                int k;
+                for (k = 0; k < currentIndex; k++)
+                {
+                    if (C->rowIndex[k] == rowA && C->colIndex[k] == colB){
+                        C->values[k] += product;
+                        found = 1;
+                        break;
+                    }
+                }
+                if (!found)
+                {
+                    C->values[currentIndex] = product;
+                    C->rowIndex[currentIndex] = rowA;
+                    C->colIndex[currentIndex] = colB;
+                    currentIndex++;
+                }
+            }
+        }
+    }
+    C->nonZeroCount = currentIndex;
+}
+
+void generate(SparseMatrix* matrix, int rows, int cols, int nonZeroCount){
+    matrix->values = (int*)malloc(sizeof(int) * nonZeroCount);
+    matrix->rowIndex = (int*)malloc(sizeof(int) * nonZeroCount);
+    matrix->colIndex = (int*)malloc(sizeof(int) * nonZeroCount);   
+    matrix->nonZeroCount = nonZeroCount;
+    int i;
+    for (i = 0; i < nonZeroCount; i++)
+    {
+        matrix->rowIndex[i] = rand() % rows;
+        matrix->colIndex[i] = rand() % cols;
+        matrix->values[i] = rand() %100;
+    }
+}
+
+void matmulNEON(float** A, float** B, float** C, int n) {
+    for (int i = 0; i < n; i++) {
+        for (int j = 0; j < n; j++) {
+            float32x4_t vecC = vmovq_n_f32(0); 
+            for (int k = 0; k < n; k += 4) {  
+                float32x4_t vecA = vld1q_f32(&A[i][k]);
+                float32x4_t vecB = vld1q_f32(&B[k][j]);
+                vecC = vmlaq_f32(vecC, vecA, vecB);  
+            }
+            C[i][j] = vgetq_lane_f32(vecC, 0) + vgetq_lane_f32(vecC, 1) +
+                       vgetq_lane_f32(vecC, 2) + vgetq_lane_f32(vecC, 3);
+        }
+    }
+}
+
+void free_matrix(SparseMatrix* matrix) {
+    free(matrix->values);
+    free(matrix->rowIndex);
+    free(matrix->colIndex);
+}
+
+int main() {
+    srand(time(NULL));
+
+    // int i;
+    int rowsA = 1000;
+    int rowsB = 2000;
+    int colsB = 1000;
+    int nonZeroCountA = 10000;
+    int nonZeroCountB = 10000;
+
+    SparseMatrix A, B;
+    generate(&A, rowsA, rowsB, nonZeroCountA);
+    generate(&B, rowsB, colsB, nonZeroCountB);
+
+    // for (i = 0; i < A.nonZeroCount; ++i) {
+    //     printf("A[%d][%d] = %d\n", A.rowIndex[i], A.colIndex[i], A.values[i]);
+    // }
+
+    // for (i = 0; i < B.nonZeroCount; ++i) {
+    //     printf("B[%d][%d] = %d\n", B.rowIndex[i], B.colIndex[i], B.values[i]);
+    // }
+    
+    int i;
+    float** matrixA = (float**)malloc(rowsA * rowsB * sizeof(float*));
+    float** matrixB = (float**)malloc(rowsB * colsB * sizeof(float*));
+    float** matrixC = (float**)malloc(rowsA * colsB * sizeof(float*));
+    for (i = 0; i < nonZeroCountA; i++)
+    {
+        int row = A.rowIndex[i];
+        int col = A.colIndex[i];
+        matrixA[row][col] = A.values[i];
+    }
+    for (i = 0; i < nonZeroCountB; i++)
+    {
+        int row = B.rowIndex[i];
+        int col = B.colIndex[i];
+        matrixA[row][col] = B.values[i];
+    }
+    
+
+    
+    SparseMatrix C;
+    C.values = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.rowIndex = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.colIndex = (int*)malloc(A.nonZeroCount * B.nonZeroCount * sizeof(int));
+    C.nonZeroCount = 0;
+
+    clock_t start_time1 = clock();
+    sparseMatmul(&A, &B, &C);
+    clock_t end_time1 = clock();
+    double time1 = (double)(end_time1 - start_time1) / CLOCKS_PER_SEC;
+    printf("sparseMatrix %f\n", time1);
+
+    clock_t start_time2 = clock();
+    matmulNEON(matrixA, matrixB, matrixC, rowsA * colsB)
+    clock_t end_time2 = clock();
+    double time2 = (double)(end_time2 - start_time2) / CLOCKS_PER_SEC;
+    printf("sparseMatrix %f\n", time2);
+
+    
+    // for (i = 0; i < C.nonZeroCount; ++i) {
+    //     printf("C[%d][%d] = %d\n", C.rowIndex[i], C.colIndex[i], C.values[i]);
+    // }
+
+    free_matrix(&A);
+    free_matrix(&B);
+    free_matrix(&C);
+    return 0;
+}