Page 1 of 1

Issues with svSolver

Posted: Thu Dec 11, 2014 6:25 am
by firemouth
Hi,

I have some issues attempting to run simulations using svSolver: I followed the instructions present in the Guide, and thus I created the files bct.dat, numstart.dat, geombc.dat.1, restart.0.1, solver.inp.
Unfortunately, when I click 'Run Simulation' I obtain the following error message


couldn't execute "/Applications/SimVascular.app/Contents/Resources/Bin/mpiexec": no such file or directory
couldn't execute "/Applications/SimVascular.app/Contents/Resources/Bin/mpiexec": no such file or directory
while executing
"exec "/Applications/SimVascular.app/Contents/Resources/Bin/mpiexec" -wdir "/Users/alessandro/Desktop/SimVascular/cylinder/Simulation-files" -np -env F..."
("eval" body line 1)
invoked from within
"eval exec \"$MPIEXEC\" -wdir \"$fullrundir\" $npflag -env FLOWSOLVER_CONFIG \"$FLOWSOLVER_CONFIG\" \"$SOLVER\" \"$fullrundir/solver.inp\" \"$stdout_fi..."
(procedure "guiRUNSOLVERlaunchSolver" line 201)
invoked from within
"guiRUNSOLVERlaunchSolver localhost"
invoked from within
".guiCV.tframe3.tpanedwindow4.tframe6.tpanedwindow0.tframe2.frame2.frame3.notebook6.tframe12.notebook0.tframe5.notebook1.tframe8.tframe1.tframe3.tframe..."
invoked from within
".guiCV.tframe3.tpanedwindow4.tframe6.tpanedwindow0.tframe2.frame2.frame3.notebook6.tframe12.notebook0.tframe5.notebook1.tframe8.tframe1.tframe3.tframe..."
(command bound to event)


Apparently, there are some problems with 'mpiexec' or something... what could I do? Maybe I missed something during the installation of the program? By the way, I'm running it with Mac OS X Yosemite.

Thanks in advance for your answers.
Alessandro

Re: Issues with svSolver

Posted: Thu Dec 11, 2014 1:45 pm
by jameson
Hello Alessandro,

I am sorry you are having trouble with the svSolver. This is indeed a bug in the simvascular gui. Thank you for reporting your issue so we can fix it in an upcoming release!! We very much appreciate feedback and bug reports.

The cause of your issue is that the GUI is trying to run mpiexec from /Applications/SimVascular.app/Contents/Resources/Bin/mpiexec but that is not where it is located. We will be fix this in the next beta release. However in the meantime, I have two work arounds that you can try.
1. Manually fix the mpiexec path in the gui (requires a change to an internal file)
2. Run svSolver from the terminal using either the mpiexec packaged with simvascular (or using your own mpiexec).

Both of these options are at your own risk. They involve modifying SimVascular files, or installing additional packages. If you wish to wait, you can subscribe to our mailing list https://simtk.org/mailman/listinfo/simvascular-news to be notified when a new version is available.

Option 1 Fix the path to mpiexec manually for the GUI.
The location of the external programs that run from the gui are located in /Applications/SimVascular.app/Contents/Resources/Tcl/externals_configure.tcl, you'll need to modify this file to run simulations from the gui.
1. Make a back up of this file.
2. Open the file using a text editor, such as TextEdit, TextWrangler, or Sublime Text and change line 6 to

Code: Select all

set gExternalPrograms(mpiexec) [file join $homedir mpiexec-mpich-gcc46]
Option 2 Run svSolver from the command line.
This option will allow you to run svSolver from the terminal without the need to open the gui. You may wish to install your own copy of mpich (which includes mpiexec)
1. Install mpich2 (Optional)
This package is available through MacPorts (https://www.macports.org). I personally use the package 'mpich-gcc46-fortran' from macports to run simulations on mac. If you do not wish to install your own copy of mpiexec you can use mpiexec that was packaged with simvascular, which is located at /Applications/SimVascular.app/Contents/Resources/mpiexec-mpich-gcc46.
2. Setup the svSolver to be run from the command line with the following command (from the terminal):

Code: Select all

sudo bash /Applications/SimVascular.app/Contents/Resources/setup-symlinks.sh
3. Follow the instructions on http://simvascular.ucsd.edu/docsFlowSolver.html to run svsolver from the command line. (Note 'mysolver' referred to in the documentation has been changed to 'svsolver')

-Jameson

Re: Issues with svSolver

Posted: Fri Dec 12, 2014 10:33 am
by firemouth
Hi Jameson,

thank you very much for your help, I really appreciate it.

For the time being, I have only tried Option 1: modifying the internal file, as you suggested, the error message disappears, but still I'm not able to run the simulation, since I obtain this log file


Start running solver...


exec "/Applications/SimVascular.app/Contents/Resources/mpiexec-mpich-gcc46" -wdir "/Users/alessandro/Desktop/SimVascular/cylinder/Simulation-files" -np -env FLOWSOLVER_CONFIG "/Applications/SimVascular.app/Contents/Resources/Bin" "/Applications/SimVascular.app/Contents/Resources/Bin/svsolver" "/Users/alessandro/Desktop/SimVascular/cylinder/Simulation-files/solver.inp" "/Users/alessandro/Desktop/SimVascular/cylinder/Log-files/alessandro-stdout_stderr-1418403702" >>& /Users/alessandro/Desktop/SimVascular/cylinder/Log-files/alessandro-1418403702.solver.log.txt &


[mpiexec@MacBook-Pro-di-Alessandro.local] HYD_pmcd_pmi_alloc_pg_scratch (pm/pmiserv/pmiserv_utils.c:527): assert (pg->pg_process_count * sizeof(struct HYD_pmcd_pmi_ecount)) failed
[mpiexec@MacBook-Pro-di-Alessandro.local] HYD_pmci_launch_procs (pm/pmiserv/pmiserv_pmci.c:108): error allocating pg scratch space
mpiexec-mpich-gcc46(1534,0x7fff7151c300) malloc: *** error for object 0xd189c88949c18945: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug



Even if I run the simulation from the terminal, it still seems there's some kind of issue with the packaged mpiexec. For this reason I'm now going to follow your second alternative and install a copy of mpich through MacPorts. I will give you feedback about that, hopefully it will work!

Thank you again, have a good day!
Alessandro

Re: Issues with svSolver

Posted: Fri Dec 12, 2014 5:50 pm
by jameson
Alessandro,

We'll look into this issue for our next release as well.

Thanks for your feedback!
-Jameson