--- /dev/null
+//**************************************************************************
+// Multi-threaded Matrix Multiply benchmark
+//--------------------------------------------------------------------------
+// TA : Christopher Celio
+// Student:
+//
+//
+// This benchmark multiplies two 2-D arrays together and writes the results to
+// a third vector. The input data (and reference data) should be generated
+// using the matmul_gendata.pl perl script and dumped to a file named
+// dataset.h.
+
+
+// print out arrays, etc.
+//#define DEBUG
+
+//--------------------------------------------------------------------------
+// Includes
+
+#include <string.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+
+//--------------------------------------------------------------------------
+// Input/Reference Data
+
+typedef float data_t;
+#include "dataset.h"
+
+
+//--------------------------------------------------------------------------
+// Basic Utilities and Multi-thread Support
+
+__thread unsigned long coreid;
+unsigned long ncores;
+
+#include "util.h"
+
+#define stringify_1(s) #s
+#define stringify(s) stringify_1(s)
+#define stats(code) do { \
+ unsigned long _c = -rdcycle(), _i = -rdinstret(); \
+ code; \
+ _c += rdcycle(), _i += rdinstret(); \
+ if (coreid == 0) \
+ printf("%s: %ld cycles, %ld.%ld cycles/iter, %ld.%ld CPI\n", \
+ stringify(code), _c, _c/DIM_SIZE/DIM_SIZE/DIM_SIZE, 10*_c/DIM_SIZE/DIM_SIZE/DIM_SIZE%10, _c/_i, 10*_c/_i%10); \
+ } while(0)
+
+
+//--------------------------------------------------------------------------
+// Helper functions
+
+void printArray( char name[], int n, data_t arr[] )
+{
+ int i;
+ if (coreid != 0)
+ return;
+
+ printf( " %10s :", name );
+ for ( i = 0; i < n; i++ )
+ printf( " %3ld ", (long) arr[i] );
+ printf( "\n" );
+}
+
+void __attribute__((noinline)) verify(size_t n, const data_t* test, const data_t* correct)
+{
+ if (coreid != 0)
+ return;
+
+ size_t i;
+ for (i = 0; i < n; i++)
+ {
+ if (test[i] != correct[i])
+ {
+ printf("FAILED test[%d]= %3ld, correct[%d]= %3ld\n",
+ i, (long)test[i], i, (long)correct[i]);
+ exit(-1);
+ }
+ }
+
+ return;
+}
+
+void matrix_sub(int size, data_t A[], data_t B[], data_t C[]) {
+ if (coreid != 0)
+ return;
+
+ for(int i = 0; i < size; i++){
+ C[i] = A[i] + B[i];
+ }
+}
+
+void matrix_add(int size, data_t A[], data_t B[], data_t C[]) {
+ if (coreid != 0)
+ return;
+
+ for(int i = 0; i < size; i++){
+ C[i] = A[i] - B[i];
+ }
+}
+
+void strassen_mult(int dime, const data_t sA[], const data_t sB[], data_t sC[]) {
+
+ if (coreid != 0)
+ return;
+
+ int height, width;
+ int sub_size = dime*dime/4;
+
+// data_t A_11[sub_size], B_11[sub_size], C_11[sub_size],
+// A_12[sub_size], B_12[sub_size], C_12[sub_size],
+// A_21[sub_size], B_21[sub_size], C_21[sub_size],
+// A_22[sub_size], B_22[sub_size], C_22[sub_size];
+
+ data_t *A_11 = malloc(sub_size*sizeof(data_t));
+ data_t *A_12 = malloc(sub_size*sizeof(data_t));
+ data_t *A_21 = malloc(sub_size*sizeof(data_t));
+ data_t *A_22 = malloc(sub_size*sizeof(data_t));
+ data_t *B_11 = malloc(sub_size*sizeof(data_t));
+ data_t *B_12 = malloc(sub_size*sizeof(data_t));
+ data_t *B_21 = malloc(sub_size*sizeof(data_t));
+ data_t *B_22 = malloc(sub_size*sizeof(data_t));
+
+ for(height=0; height < dime/2; height++) {
+ for(width= 0; width < dime/2; width++) {
+ A_11[width+(height*dime/2)] = sA[width + height*dime];
+ B_11[width+(height*dime/2)] = sB[width + height*dime];
+
+ A_12[width+(height*dime/2)] = sA[dime/2 + width + height*dime];
+ B_12[width+(height*dime/2)] = sB[dime/2 + width + height*dime];
+
+ A_21[width+(height*dime/2)] = sA[(dime*dime)/2 + width + height*dime];
+ B_21[width+(height*dime/2)] = sB[(dime*dime)/2 + width + height*dime];
+
+ A_22[width+(height*dime/2)] = sA[(dime*dime)/2 + dime/2 + width + height*dime];
+ B_22[width+(height*dime/2)] = sB[(dime*dime)/2 + dime/2 + width + height*dime];
+ }
+ }
+
+// data_t H_1[sub_size], H_2[sub_size], H_3[sub_size], H_4[sub_size], H_5[sub_size],
+// H_6[sub_size], H_7[sub_size], H_8[sub_size], H_9[sub_size], H_10[sub_size],
+// H_11[sub_size], H_12[sub_size], H_13[sub_size], H_14[sub_size],
+// H_15[sub_size], H_16[sub_size], H_17[sub_size], H_18[sub_size];
+
+ data_t *H_1 = malloc(sub_size*sizeof(data_t));
+ data_t *H_2 = malloc(sub_size*sizeof(data_t));
+ data_t *H_3 = malloc(sub_size*sizeof(data_t));
+ data_t *H_4 = malloc(sub_size*sizeof(data_t));
+ data_t *H_5 = malloc(sub_size*sizeof(data_t));
+ data_t *H_6 = malloc(sub_size*sizeof(data_t));
+ data_t *H_7 = malloc(sub_size*sizeof(data_t));
+ data_t *H_8 = malloc(sub_size*sizeof(data_t));
+ data_t *H_9 = malloc(sub_size*sizeof(data_t));
+ data_t *H_10 = malloc(sub_size*sizeof(data_t));
+
+ matrix_add(sub_size, A_11, A_22, H_1); //Helper1
+ matrix_add(sub_size, B_11, B_22, H_2); //Helper2
+ matrix_add(sub_size, A_21, A_22, H_3); //Helper3
+ matrix_sub(sub_size, B_12, B_22, H_4); //Helper4
+ matrix_sub(sub_size, B_21, B_11, H_5); //Helper5
+ matrix_add(sub_size, A_11, A_12, H_6); //Helper6
+ matrix_sub(sub_size, A_21, A_11, H_7); //Helper7
+ matrix_add(sub_size, B_11, B_12, H_8); //Helper8
+ matrix_sub(sub_size, A_12, A_22, H_9); //Helper9
+ matrix_add(sub_size, B_21, B_22, H_10); //Helper10
+
+ free(A_12);
+ free(A_21);
+ free(B_12);
+ free(B_21);
+
+ A_12 = NULL;
+ A_21 = NULL;
+ B_12 = NULL;
+ B_21 = NULL;
+
+// data_t M_1[sub_size], M_2[sub_size], M_3[sub_size], M_4[sub_size],
+// M_5[sub_size], M_6[sub_size], M_7[sub_size];
+
+ data_t *M_1 = malloc(sub_size*sizeof(data_t));
+ data_t *M_2 = malloc(sub_size*sizeof(data_t));
+ data_t *M_3 = malloc(sub_size*sizeof(data_t));
+ data_t *M_4 = malloc(sub_size*sizeof(data_t));
+ data_t *M_5 = malloc(sub_size*sizeof(data_t));
+ data_t *M_6 = malloc(sub_size*sizeof(data_t));
+ data_t *M_7 = malloc(sub_size*sizeof(data_t));
+
+ if (sub_size == 1) {
+ M_1[0] = H_1[0]*H_2[0];
+ M_2[0] = H_3[0]*B_11[0];
+ M_3[0] = A_11[0]*H_4[0];
+ M_4[0] = A_22[0]*H_5[0];
+ M_5[0] = H_6[0]*B_22[0];
+ M_6[0] = H_7[0]*H_8[0];
+ M_7[0] = H_9[0]*H_10[0];
+ } else {
+ strassen_mult(dime/2, H_1, H_2, M_1);
+ strassen_mult(dime/2, H_3, B_11, M_2);
+ strassen_mult(dime/2, A_11, H_4, M_3);
+ strassen_mult(dime/2, A_22, H_5, M_4);
+ strassen_mult(dime/2, H_6, B_22, M_5);
+ strassen_mult(dime/2, H_7, H_8, M_6);
+ strassen_mult(dime/2, H_9, H_10, M_7);
+ }
+
+ free(A_11);
+ free(A_22);
+ free(B_11);
+ free(B_22);
+
+ A_11 = NULL;
+ A_22 = NULL;
+ B_11 = NULL;
+ B_22 = NULL;
+
+ free(H_1);
+ free(H_2);
+ free(H_3);
+ free(H_4);
+ free(H_5);
+ free(H_6);
+ free(H_7);
+ free(H_8);
+ free(H_9);
+ free(H_10);
+
+ H_1 = NULL;
+ H_2 = NULL;
+ H_3 = NULL;
+ H_4 = NULL;
+ H_5 = NULL;
+ H_6 = NULL;
+ H_7 = NULL;
+ H_8 = NULL;
+ H_9 = NULL;
+ H_10 = NULL;
+
+ data_t *H_11 = malloc(sub_size*sizeof(data_t));
+ data_t *H_12 = malloc(sub_size*sizeof(data_t));
+ data_t *H_13 = malloc(sub_size*sizeof(data_t));
+ data_t *H_14 = malloc(sub_size*sizeof(data_t));
+
+ data_t *C_11 = malloc(sub_size*sizeof(data_t));
+ data_t *C_12 = malloc(sub_size*sizeof(data_t));
+ data_t *C_21 = malloc(sub_size*sizeof(data_t));
+ data_t *C_22 = malloc(sub_size*sizeof(data_t));
+
+ matrix_add(sub_size, M_1, M_4, H_11);
+ matrix_add(sub_size, M_5, M_7, H_12);
+ matrix_sub(sub_size, H_11, H_12, C_11);
+
+ matrix_add(sub_size, M_3, M_5, C_12);
+
+ matrix_add(sub_size, M_2, M_4, C_21);
+
+ matrix_sub(sub_size, M_1, M_2, H_13);
+ matrix_add(sub_size, M_3, M_6, H_14);
+ matrix_add(sub_size, H_13, H_14, C_22);
+
+ free(H_11);
+ free(H_12);
+ free(H_13);
+ free(H_14);
+
+ H_11 = NULL;
+ H_12 = NULL;
+ H_13 = NULL;
+ H_14 = NULL;
+
+
+ for(height=0; height < dime/2; height++) {
+ for(width= 0; width < dime/2; width++) {
+ sC[width + height*dime] = C_11[width+(height*dime/2)];
+ sC[dime/2 + width + height*dime] = C_12[width+(height*dime/2)];
+ sC[(dime*dime)/2 + width + height*dime] = C_21[width+(height*dime/2)];
+ sC[(dime*dime)/2 + dime/2 + width + height*dime] = C_22[width+(height*dime/2)];
+ }
+ }
+
+ free(C_11);
+ free(C_12);
+ free(C_21);
+ free(C_22);
+
+ C_11 = NULL;
+ C_12 = NULL;
+ C_21 = NULL;
+ C_22 = NULL;
+
+}
+
+//--------------------------------------------------------------------------
+// matmul function
+
+// single-thread, naive version
+void __attribute__((noinline)) matmul_naive(const int lda, const data_t A[], const data_t B[], data_t C[] )
+{
+ int i, j, k;
+
+ if (coreid > 0)
+ return;
+
+ for ( i = 0; i < lda; i++ )
+ for ( j = 0; j < lda; j++ )
+ {
+ for ( k = 0; k < lda; k++ )
+ {
+ C[i + j*lda] += A[j*lda + k] * B[k*lda + i];
+ }
+ }
+
+}
+
+
+
+void __attribute__((noinline)) matmul(const int lda, const data_t A[], const data_t B[], data_t C[] )
+{
+
+ // ***************************** //
+ // **** ADD YOUR CODE HERE ***** //
+ // ***************************** //
+ //
+ // feel free to make a separate function for MI and MSI versions.
+
+ if (coreid > 0)
+ return;
+
+ strassen_mult(lda, A, B, C);
+
+}
+
+//--------------------------------------------------------------------------
+// Main
+//
+// all threads start executing thread_entry(). Use their "coreid" to
+// differentiate between threads (each thread is running on a separate core).
+
+void thread_entry(int cid, int nc)
+{
+ coreid = cid;
+ ncores = nc;
+
+ // static allocates data in the binary, which is visible to both threads
+ static data_t results_data[ARRAY_SIZE];
+
+
+// // Execute the provided, naive matmul
+// barrier();
+// stats(matmul_naive(DIM_SIZE, input1_data, input2_data, results_data); barrier());
+//
+//
+// // verify
+// verify(ARRAY_SIZE, results_data, verify_data);
+//
+// // clear results from the first trial
+// size_t i;
+// if (coreid == 0)
+// for (i=0; i < ARRAY_SIZE; i++)
+// results_data[i] = 0;
+// barrier();
+
+
+ // Execute your faster matmul
+ barrier();
+ stats(matmul(DIM_SIZE, input1_data, input2_data, results_data); barrier());
+
+#ifdef DEBUG
+ printArray("results:", ARRAY_SIZE, results_data);
+ printArray("verify :", ARRAY_SIZE, verify_data);
+#endif
+
+ // verify
+ verify(ARRAY_SIZE, results_data, verify_data);
+ barrier();
+
+ exit(0);
+}
+
projects / riscv-tests.git / blobdiff