Hi guys. I tried to run current hama on some node cluster. I was worried about disk IO performance problems, but it seems work well better than expected. In this step I would like to recommend that we'll attempt to handle/optimize only dense matrix with blocking/tiling algorithms on the map/reduce.
The sparse matrix operations will be quite different - this is a next step. Any ideas? -- Best regards, Edward J. Yoon @ NHN, corp. [EMAIL PROTECTED] http://blog.udanax.org
