}
public void run() {
- int localresults[][];
+ double localresults[][];
atomic {
//compute the results
- localresults=new int[1+x1-x0][1+y1-y0];
-
+ localresults=new double[1+x1-x0][1+y1-y0];
+ double la[][]=mmul.a;
+ double lbtranspose[][]=mmul.b;
+ double lc[][]=mmul.c;
+ int M=mmul.M;
+
+ //Use b transpose for cache performance
for(int i = x0; i<= x1; i++){
- int a[]=mmul.a[i];
- int b[][]=mmul.b;
- int M=mmul.M;
+ double a[]=la[i];
for (int j = y0; j <= y1; j++) {
- int innerProduct=0;
+ double innerProduct=0;
+ double b[] = lbtranspose[j];
for(int k = 0; k < M; k++) {
- innerProduct += a[k] * b[k][j];
+ innerProduct += a[k] *b[k];
}
localresults[i-x0][j-y0]=innerProduct;
}
}
}
+
atomic {
//write the results
for(int i=x0;i<=x1;i++) {
- int c[]=mmul.c[i];
+ double c[]=mmul.c[i];
for(int j=y0;j<=y1;j++) {
c[j]=localresults[i-x0][j-y0];
}
}
public static void main(String[] args) {
- int mid1 = (128<<24)|(195<<16)|(175<<8)|70;
+<<<<<<< MatrixMultiply.java
+ int mid1 = (128<<24)|(195<<16)|(175<<8)|69;
int mid2 = (128<<24)|(195<<16)|(175<<8)|69;
int mid3 = (128<<24)|(195<<16)|(175<<8)|71;
- int NUM_THREADS = 2;
- int i, j, p, q, r;
+ int NUM_THREADS = 1;
+=======
+ int NUM_THREADS = 4;
+ int[] mid = new int[NUM_THREADS];
+ mid[0] = (128<<24)|(195<<16)|(175<<8)|69;
+ mid[1] = (128<<24)|(195<<16)|(175<<8)|73;
+ mid[2] = (128<<24)|(195<<16)|(175<<8)|78;
+ mid[3] = (128<<24)|(195<<16)|(175<<8)|79;
+ //int mid1 = (128<<24)|(195<<16)|(175<<8)|69;
+ //int mid2 = (128<<24)|(195<<16)|(175<<8)|73;
+>>>>>>> 1.19
+ int p, q, r;
MatrixMultiply[] mm;
MatrixMultiply tmp;
MMul matrix;
atomic {
- matrix = global new MMul(70, 70, 70);
+ matrix = global new MMul(400, 400, 400);
matrix.setValues();
+ matrix.transpose();
}
atomic{
mm = global new MatrixMultiply[NUM_THREADS];
}
- // Currently it is a 70 X 70 matrix divided into 4 blocks
atomic {
- mm[0] = global new MatrixMultiply(matrix,0,0,69,35);
- mm[1] = global new MatrixMultiply(matrix,0,36,69,69);
+ mm[0] = global new MatrixMultiply(matrix,0,0,200,200);
+ mm[1] = global new MatrixMultiply(matrix,0,201,200,399);
+ mm[2] = global new MatrixMultiply(matrix,201,0,399,200);
+ mm[3] = global new MatrixMultiply(matrix,201,201,399,399);
}
atomic {
System.printInt(r);
System.printString("\n");
- //Print Matrices to be multiplied
- /*
- System.printString("a =\n");
- for (i = 0; i < p; i++) {
- for (j = 0; j < q; j++) {
- atomic {
- val = matrix.a[i][j];
- }
- System.printString(" " + val);
- }
- System.printString("\n");
- }
- System.printString("\n");
-
- System.printString("b =\n");
- for (i = 0; i < q; i++) {
- for (j = 0; j < r; j++) {
- atomic {
- val = matrix.b[i][j];
- }
- System.printString(" " + val);
- }
- System.printString("\n");
- }
- System.printString("\n");
- */
-
// start a thread to compute each c[l,n]
- for (i = 0; i < NUM_THREADS; i++) {
+ for (int i = 0; i < NUM_THREADS; i++) {
atomic {
tmp = mm[i];
}
- tmp.start(mid1);
+ tmp.start(mid[i]);
}
// wait for them to finish
- for (i = 0; i < NUM_THREADS; i++) {
+ for (int i = 0; i < NUM_THREADS; i++) {
atomic {
tmp = mm[i];
}
// print out the result of the matrix multiply
System.printString("Starting\n");
System.printString("Matrix Product c =\n");
- StringBuffer sb = new StringBuffer("");
- int val;
+ double val;
atomic {
- for (i = 0; i < p; i++) {
- int c[]=matrix.c[i];
- for (j = 0; j < r; j++) {
+ for (int i = 0; i < p; i++) {
+ double c[]=matrix.c[i];
+ for (int j = 0; j < r; j++) {
val = c[j];
- sb.append(" ");
- sb.append((new Integer(val)).toString());
}
- sb.append("\n");
}
}
- System.printString(sb.toString());
System.printString("Finished\n");
}
}
public class MMul{
public int L, M, N;
- public int[][] a;
- public int[][] b;
- public int[][] c;
+ public double[][] a;
+ public double[][] b;
+ public double[][] c;
+ public double[][] btranspose;
public MMul(int L, int M, int N) {
this.L = L;
this.M = M;
this.N = N;
- a = global new int[L][M];
- b = global new int[M][N];
- c = global new int[L][N];
+ a = global new double[L][M];
+ b = global new double[M][N];
+ c = global new double[L][N];
+ btranspose = global new double[N][M];
}
public void setValues() {
- int i;
- int j;
- for(i = 0; i < L; i++) {
- for(j = 0; j < M; j++) {
- a[i][j] = j+1;
+ for(int i = 0; i < L; i++) {
+ double ai[] = a[i];
+ for(int j = 0; j < M; j++) {
+ ai[j] = j+1;
}
}
- for(i = 0; i < M; i++) {
- for(j = 0; j < N; j++) {
- b[i][j] = j+1;
+ for(int i = 0; i < M; i++) {
+ double bi[] = b[i];
+ for(int j = 0; j < N; j++) {
+ bi[j] = j+1;
}
}
- for(i = 0; i < L; i++) {
- for(j = 0; j < N; j++) {
- c[i][j] = 0;
+ for(int i = 0; i < L; i++) {
+ double ci[] = c[i];
+ for(int j = 0; j < N; j++) {
+ ci[j] = 0;
+ }
+ }
+ for(int i = 0; i < N; i++) {
+ double btransposei[] = btranspose[i];
+ for(int j = 0; j < M; j++) {
+ btransposei[j] = 0;
}
}
}
-}
+ public void transpose() {
+ for(int row = 0; row < M; row++) {
+ double brow[] = b[row];
+ for(int col = 0; col < N; col++) {
+ btranspose[col][row] = brow[col];
+ }
+ }
+ }
+}