DMP Distributed Matrix Processor.

It uses akka cluster for message passing and cluster managment/routing etc.

To try out follow these steps.

• Setup: For quick trying out purposes and minimal configuration a few nodes can be started locally.

  $ sbt dist

  • After this copy the generated dist in target/dist to your convenient location and replace bin/* with scripts/*

  • Then start four different nodes by doing this in four different terminals.

    $ bin/start kernel.Backend 0

    $ bin/start kernel.Backend 1

    $ bin/start kernel.Backend 2

    $ bin/start kernel.Backend 3

• Using: Once backend nodes are running. You connect via sbt console and try multiplying example matrices.

  $ sbt> console

...... prints logo and some commands ......

 `scala> import scala.collection.mutable.ArrayBuffer`

 `scala> val A = DistributedMatrix("A", 8, 9, noOfBlocks, blockSize, ArrayBuffer(1 to 72: _*))`

 `scala> val B = DistributedMatrix("B", 8, 9, noOfBlocks, blockSize, ArrayBuffer(1 to 72: _*))`

 `scala> A.persist //Will persist the sample matrix A`

 `scala> B.persist //Similiarly will persist the sample matrix B`

 `scala> val C = A x B // this will multiply them, and it happens on the cluster not local.`

 `scala> val C = C x B // this will again multiply them.`

 `scala> C.getMatrix // To see the content of the matrix and get a local matrix from DM.`

Thanks!

Please open issues to share opinions. At the moment have no mailing lists setup etc.