//************************************************************************** // 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 #include #include //-------------------------------------------------------------------------- // 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; } //-------------------------------------------------------------------------- // 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. int m, i, j, k, iB0, iB1; data_t tempC0, tempC1, tempC2, tempC3, tempC4, tempC5, tempC6, tempC7; data_t tempA0, tempA1; if (coreid == 0){ for (m = 0; m < 2; m++){ for (j = 0; j < lda/2; j++){ for (i = 0; i < lda; i+=8){ tempC0 = C[i + j*lda]; tempC1 = C[i + j*lda+1]; tempC2 = C[i + j*lda+2]; tempC3 = C[i + j*lda+3]; tempC4 = C[i + j*lda+4]; tempC5 = C[i + j*lda+5]; tempC6 = C[i + j*lda+6]; tempC7 = C[i + j*lda+7]; iB0 = m*lda*lda/2+i; iB1 = iB0+lda; for (k = m*lda/2; k < (m+1)*lda/2; k+=2){ tempA0 = A[j*lda+k]; tempA1 = A[j*lda+k+1]; tempC0 += tempA0*B[iB0]+tempA1*B[iB1]; tempC1 += tempA0*B[iB0+1]+tempA1*B[iB1+1]; tempC2 += tempA0*B[iB0+2]+tempA1*B[iB1+2]; tempC3 += tempA0*B[iB0+3]+tempA1*B[iB1+3]; tempC4 += tempA0*B[iB0+4]+tempA1*B[iB1+4]; tempC5 += tempA0*B[iB0+5]+tempA1*B[iB1+5]; tempC6 += tempA0*B[iB0+6]+tempA1*B[iB1+6]; tempC7 += tempA0*B[iB0+7]+tempA1*B[iB1+7]; iB0 += 2*lda; iB1 += 2*lda; } C[i + j*lda] = tempC0; C[i + j*lda + 1] = tempC1; C[i + j*lda + 2] = tempC2; C[i + j*lda + 3] = tempC3; C[i + j*lda + 4] = tempC4; C[i + j*lda + 5] = tempC5; C[i + j*lda + 6] = tempC6; C[i + j*lda + 7] = tempC7; } } } } else { for (m = 2; m > 0; m--){ for (j = lda-1; j >= lda/2; j--){ for (i = lda-1; i >= 0; i-=8){ tempC0 = C[i + j*lda]; tempC1 = C[i + j*lda - 1]; tempC2 = C[i + j*lda - 2]; tempC3 = C[i + j*lda - 3]; tempC4 = C[i + j*lda - 4]; tempC5 = C[i + j*lda - 5]; tempC6 = C[i + j*lda - 6]; tempC7 = C[i + j*lda - 7]; for (k = m*lda/2-1; k >= (m-1)*lda/2; k-=2){ tempA0 = A[j*lda+k]; tempA1 = A[j*lda+k-1]; tempC0 += tempA0*B[k*lda+i]+tempA1*B[(k-1)*lda+i]; tempC1 += tempA0*B[k*lda+i-1]+tempA1*B[(k-1)*lda+i-1]; tempC2 += tempA0*B[k*lda+i-2]+tempA1*B[(k-1)*lda+i-2]; tempC3 += tempA0*B[k*lda+i-3]+tempA1*B[(k-1)*lda+i-3]; tempC4 += tempA0*B[k*lda+i-4]+tempA1*B[(k-1)*lda+i-4]; tempC5 += tempA0*B[k*lda+i-5]+tempA1*B[(k-1)*lda+i-5]; tempC6 += tempA0*B[k*lda+i-6]+tempA1*B[(k-1)*lda+i-6]; tempC7 += tempA0*B[k*lda+i-7]+tempA1*B[(k-1)*lda+i-7]; } C[i + j*lda] = tempC0; C[i + j*lda - 1] = tempC1; C[i + j*lda - 2] = tempC2; C[i + j*lda - 3] = tempC3; C[i + j*lda - 4] = tempC4; C[i + j*lda - 5] = tempC5; C[i + j*lda - 6] = tempC6; C[i + j*lda - 7] = tempC7; } } } } } //-------------------------------------------------------------------------- // 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); }