I'm not sure whether this question should be fielded to the VS or HPC forums, but I'll try here first.
We're testing the parallel debug feature from VS2010 against a Windows 2008 R2 SP2 HPC cluster. While we can get jobs to submit and run, I have a couple of issues.
First, when a break point is hit for a job running on the cluster, is there a way to get the console output window on the client submitting the job to open with the debug details? If the job is run locally, the console window opens ok. If the job is
running on the cluster, then it doesn't appear. I'm guessing this might be normal behaviour, but it would be useful to be able to get the console window to open.
Second, if debug session is submitted to the cluster for a single-threaded app, but this app spawns various other threads (up to 6), would the cluster process each spawned thread on a different core, or would they all process on the same core? The job is
being submitted with the 'per core' MPI debug setting.