Hi Ronald,
You need give more information on how your number-crunching apps work. A few ways to parallel your app:
1. Multiple runs for your application on mulitple machines. Think of a movie transcoding work, usually you do with below on one node:
TranscodeApp.exe <targetMovieFile> <OutputFile>
Now with HPC Cluster, you need start multiple instance of transcode but each targets part of the MovieFile
On Node1: TranscodeApp.exe <targetMovieFile>.<Part1> <OutputFile><Part1>
On Node2: TranscodeApp.exe <targetMovieFile>.<Part2> <OutputFile><Part2>
On NodeN: TranscodeApp.exe <targetMovieFile>.<PartN> <OutputFile><PartN>
And the last task is to merge the outputFile together.
2. If your application can run on mulitple nodes, just start the application in the cluster
If you can provide more details of your workload and how your application works, I may provide more appropriate suggestions.
Qiufang Shi
