StaMPS 4.1b Parallelisation

[Stoorm]

Hi all,
Does anyone know how to properly use the parallelization function in the new StaMPS 4.1b?
For example I have a workstation with 2 processors, 6 physical cores each (12 logical) and I have a data set consisting of 10 patches. What are the best n_cores values to choose? What types of cores should I use - physical or logical (HT)?
Аfter running stamps_mc_header (1.5) I get messages about creating the jobs for matlab. How to run the process further?
Thank you!

[David Bekaert]

Hi,

n_cores will gives n_cores different stamps commands to be run each with its own patch list.

Its not really an automatic parallelization, though an attempt to split the patch list and allow the user to launch multiple stamps jobs.

You can next run the command as outputted to the screen in different screens and/or on different machines.

Given the processing goes on patches, it is only applicable to step 1 - step 5 first part.

Once all these processes have completed, then you can run the command for second part of step 5, or you could just decide to run step 5 with the conventional stamps.m script.

Past step 5 you should just use stamps as is.

Cheers,

D.

--
You received this message because you are subscribed to the Google Groups "MAINSAR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mainsar+u...@googlegroups.com.
To post to this group, send email to mai...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Stoorm]

David,

Thank you for explanation!

понедельник, 10 сентября 2018 г., 17:47:27 UTC+3 пользователь David Bekaert написал: