Standard way of handling large mesh in parallel run
******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hi Everyone I've been trying to import a mesh generated in a third-party package and run it in the incompressible solver. First, I tried a smaller mesh with a size of 10MB and 32 cores on an HPC with 3GB ram specified for each core. It worked and I proceeded to the main larger mesh with a size of 75MB and 128 cores. However, I found it can be quite challenging to establish a simulation as I often get MPI-related errors at the very beginning. I was wondering if this is the standard way of handling large mesh files in parallel using mpirun: mpirun -np 128 IncNavierStokesSolver condition.xml grid.xml I am new to nektar++ so my question might seem trivial. Kind regards syavash
Hi Syavash, Is the mesh that you are using converted to a compatible format with nektar++? The command should be fine, however I would recommend to first type the mesh and then the conditions, so the command should look like: mpirun -np 128 IncNavierStokesSolver grid.xml condition.xml Also, another good practise when dealing with large meshes is to install nektar++ with hdf5 and run your simulations using this format for exporting the fields. I hope this helps! If you are experiencing some specific errors, feel free to share a picture with the error and the output of the solver so that we can help you more with it. Kind regards, Alexandra From: nektar-users-bounces@imperial.ac.uk <nektar-users-bounces@imperial.ac.uk> On Behalf Of Ehsan Asgari Sent: 08 March 2023 17:19 To: Nektar-users@imperial.ac.uk Subject: [Nektar-users] Standard way of handling large mesh in parallel run This email from eh.asgari@gmail.com <mailto:eh.asgari@gmail.com> originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list <https://spam.ic.ac.uk/SpamConsole/Senders.aspx> to disable email stamping for this address. Hi Everyone I've been trying to import a mesh generated in a third-party package and run it in the incompressible solver. First, I tried a smaller mesh with a size of 10MB and 32 cores on an HPC with 3GB ram specified for each core. It worked and I proceeded to the main larger mesh with a size of 75MB and 128 cores. However, I found it can be quite challenging to establish a simulation as I often get MPI-related errors at the very beginning. I was wondering if this is the standard way of handling large mesh files in parallel using mpirun: mpirun -np 128 IncNavierStokesSolver condition.xml grid.xml I am new to nektar++ so my question might seem trivial. Kind regards syavash
Hi Syavash, As Alexandra said - if you can provide the output of the solver on crash that would be extremely helpful to diagnose any issue. In terms of the mesh size - these are small meshes relative to some of meshes we have run internally. Can you provide any details on how you have generated the mesh? I.e: * The meshing software you have used and version number * The cell count and cell types i.e Tris,Quads for 2D or Tet,Prism,Hex for 3D * And your procedure to convert to a Nektar++ mesh file i.e. your process to import to a .xml Additionally the mesh file plus the session file and a description of your case would be useful - as it will help to diagnose the issue. Kind regards, James Slaughter. ________________________________ From: nektar-users-bounces@imperial.ac.uk <nektar-users-bounces@imperial.ac.uk> on behalf of Ehsan Asgari <eh.asgari@gmail.com> Sent: 08 March 2023 17:19 To: nektar-users <nektar-users@imperial.ac.uk> Subject: [Nektar-users] Standard way of handling large mesh in parallel run This email from eh.asgari@gmail.com originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list<https://spam.ic.ac.uk/SpamConsole/Senders.aspx> to disable email stamping for this address. Hi Everyone I've been trying to import a mesh generated in a third-party package and run it in the incompressible solver. First, I tried a smaller mesh with a size of 10MB and 32 cores on an HPC with 3GB ram specified for each core. It worked and I proceeded to the main larger mesh with a size of 75MB and 128 cores. However, I found it can be quite challenging to establish a simulation as I often get MPI-related errors at the very beginning. I was wondering if this is the standard way of handling large mesh files in parallel using mpirun: mpirun -np 128 IncNavierStokesSolver condition.xml grid.xml I am new to nektar++ so my question might seem trivial. Kind regards syavash
participants (3)
-
Ehsan Asgari
-
liosi.alex＠gmail.com
-
Slaughter, James W