Simulation Not Running

[David Parker]

Hi Susannah,

The LINE 0000, LINE 0001 messages refer to the line number in the svSolver input file (.svpre) stored under the project's Simulation directory.

The line number 2 in this file has a bad format

set_surface_id_vtp mesh-complete/mesh-surfaces/cap_contour g

should look something like this

set_surface_id_vtp mesh-complete/mesh-surfaces/inflow.vtp 2

There should be a number, not the character g, after the file name and the svpre program is failing because of that. And the file name does not end in .vtp either, curious.

It's not clear how the .svpre could be incorrect. Are you using the latest SimVascular version?

Cheers,
Dave

[Sita Drost]

Hi all,

I seem to have the same problem as Diwas, but on a MacBook (Mojave, 10.14.2): "create data files for simulation" works alright, so then I press "Run Simulation" (and then yes), and in the lower left corner I see "pulsatile_sim: running. 0% complete". It has been "running" for hours by now, but nothing is happening, and even though I recruited 4 processors, my CPU load is only around 2%. Does anyone know how to resolve this? Am I doing something wrong?

Thanks in advance!
Sita

[David Parker]

Hi Sita,

You should only use a single processor when running svSolver on a laptop or a workstation, it was designed to run on a cluster and does not do what you think it should (i.e. run using multiple cores) on a laptop. I am going to open a GitHub Issue on this.

Cheers,
Dave

[Sita Drost]

Hi Dave,

Thanks for your reply, great to get such fast response. In the meantime I already tried using just one processor, but unfortunately that doesn't seem to make a difference. I also downloaded the Cylinder Project, hoping that a simpler test case, with a smaller mesh might work, but no luck so far. I'm not getting any error messages though, it's just that nothing is happening. Are there any tests I could run to find out what I'm doing wrong?

Thanks!
Sita

[Sita Drost]

Cool, when I run the simulation from the command line it does work! Apparently something goes wrong when running a simulation from the GUI, but I'm glad I can at least run things from the command line, will stick to that for now.

Cheers,
Sita

[David Parker]

I just tested on OSX and there is indeed a problem running svSolver. I will have a look.

Cheers,
Dave

[Sita Drost]

By the way, from the command line I seem to be able run the svSolver on multiple cores in parallel on my MacBook alright. Great, that'll save me tons of time!

[Sita Drost]

Ah right, so that's what you meant by it does not do what you expect on a laptop... I've 4 instances of svSolver running in parallel, but it's barely any faster than running the same job on a single core. Too bad...

[Daniel Romary]

Hi all,

Just wondering if there has been any progress on this issue? I am having the same problem where the simulation sits at 0% complete forever on my Mac.

If the workaround is to run it from the command line, could someone tell me what code I need to enter to run it from the command line?

Thanks so much!

-Daniel

[David Parker]

Hi Daniel,

There are still some problems with SimVascular finding the correct mpiexec version on MacOS, I hope to get a fix in for this soon.

In the mean time here are some things to check (also see the GitGub Issue https://github.com/SimVascular/svSolver/issues/36).

1) svsolver is expecting to use mpich2 on MacOS so make sure the mpiexec you are using is from mpich2.

2) set the mpiexec path to /usr/local/bin/mpiexec (or wherever the mpich mpi installation is) under the SimVascular Preferences menu
located under the SimVascular icon on the Mac toolbar.

You can run the solver on the command line by changing directories to your project's Simulations / <JOBNAME> directory (<JOBNAME> is the name of the job you typed in when creating the simulation from SimVascular). In that directory you should see: numstart.dat, restart.0.1, solver.inp and a .svpre files. To run the solver type

mpiexec -n<NPROC> /usr/local/sv/svsolver/2019-01-19/svsolver solver.inp

That will execute svsolver with <NPROC> processors.

Cheers,
Dave