Forum
Important Notice for New User Registrations
To combat an increasing number of spam and bot registrations, we now manually approve all new user registrations. While this may cause a delay until your account is approved, this step is essential to ensure the quality and security of this forum.
To help us verify your registration as legitimate, please use a clear name as user name or an official email address (such as a work, university, or similar address). If you’re concerned that we may not recognize your registration as non-spam, feel free to email us at with a request to approve your username.
Assistance with Parallelization on Cluster for SIL Interface VAWT Simulations
Good afternoon,
I am currently modeling a VAWT using the SIL Interface. To speed up my simulations, I would like to run them on a 40-core cluster. I noticed that the simulations seem to parallelize well on my local machine (8 cores), but when running them on the cluster, htop shows that I am not fully utilizing the 40 cores as expected.
Could you please advise on how to ensure that my simulation properly utilizes all available cores on the cluster?
Thank you in advance for your help.
Best regards,
Serge El Asmar
Hi Serge,
When you run a single QBlade simulation, some parts already use OpenMP to run in parallel. But not everything is parallelized the same way. For example, the structural solver only uses one core. To make better use of a cluster, you should run several QBlade SIL instances at the same time using a script (like in Python or Matlab). This lets you run multiple simulations in parallel. As a rough guide, performance is usually best when you use about half the available cores, but this can vary depending on your system.
Best regards,
David
