Hi, I just made a pseudo code for this issue. I'm sure it will show better performance than any M/R version of mat-mat mult on Hadoop.
Following is the sample code you could reference when porting to BSP version, http://svn.apache.org/repos/asf/incubator/hama/trunk/src/test/org/apache/hama/bsp/BSPPeerTest.java Does anyone have a interesting or want to implement this? On Fri, Feb 12, 2010 at 11:20 AM, Edward J. Yoon (JIRA) <[email protected]> wrote: > > [ > https://issues.apache.org/jira/browse/HAMA-221?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12832796#action_12832796 > ] > > Edward J. Yoon commented on HAMA-221: > ------------------------------------- > > Here's my multi threaded version of mat-mat mult. > > In this example, > - Thread.join() means a barrier. > - The global storage (int[][] out) was used. (Of course, you can also split > the summation into n-supersteps.) > > ---- > class MatMult extends Thread { > static int m1[][]; > static int m2[][]; > static int out[][]; > static int n = 2; > int row; > > MatMult(int i) { > row = i; > this.start(); > } > > public void run() { > int i, j; > for (i = 0; i < n; i++) { > out[row][i] = 0; > > for (j = 0; j < n; j++) { > out[row][i] = out[row][i] + m1[row][j] * m2[j][i]; > } > } > } > > public static void main(String args[]) throws Exception { > int i, j; > int n = 2; > m1 = new int[][] { { 1, 2 }, { 3, 4 } }; > m2 = new int[][] { { 1, 2 }, { 3, 4 } }; > out = new int[n][n]; > > MatMult mat[] = new MatMult[n]; > for (i = 0; i < n; i++) { > mat[i] = new MatMult(i); > } > > for (i = 0; i < n; i++) { > mat[i].join(); > } > > System.out.println("OUTPUT :"); > for (i = 0; i < n; i++) { > for (j = 0; j < n; j++) { > System.out.print(out[i][j] + "\t"); > } > System.out.println(); > } > } > } > > >> A BSP matrix multiplication implementation. >> ------------------------------------------- >> >> Key: HAMA-221 >> URL: https://issues.apache.org/jira/browse/HAMA-221 >> Project: Hama >> Issue Type: Improvement >> Components: bsp, matrix >> Affects Versions: 0.2.0 >> Reporter: Edward J. Yoon >> Fix For: 0.2.0 >> >> >> Using the BSP model, we could improve the performance of mat-mat >> multiplication. Let's evaluation it out on this issue. > > -- > This message is automatically generated by JIRA. > - > You can reply to this email to add a comment to the issue online. > > -- Best Regards, Edward J. Yoon @ NHN, corp. [email protected] http://blog.udanax.org
