-
Notifications
You must be signed in to change notification settings - Fork 859
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Bug during running a code in GPU when Resolution is increased but runs well in CPU #12778
Comments
@atanuchaudhury The error appears to be a linking issue between the application and MPI. Typically this should be mitigated on the application side. Have you reached out to Cylinder3d community?
|
I used it but the error now it gives: |
Actually I have extended the along the Y axis which is upto 2 meter and rest are the same. So do I need to change the prepareGeometry setup? What changes should I make? Please reply |
I think you are mixing two versions of Open MPI. |
I am giving you the details what I have done till now. my openmpi version: First the changes that I made in the Cylinder3D code:
} Now I have made a new stl file in which I extended the Y axis to 2000 mm from the previous 410 mm and rest remains the same. So with Resolution, N = 10, it runs smoothly but as I increase the resolution say N = 20, 50 , 100 etc. and run it in GPU, the simulation gets abborted and shows error. I posted it to the Openlb forum but they are telling that its not the issue of mpirun. Please help me to solve it. The error is given below :
|
Unfortunately, there is little we can help with.
|
@atanuchaudhury please run the following commands and post the outputs:
|
Ityped those commands and got these messages: Report bugs to http://www.open-mpi.org/community/help/ Report bugs to http://www.open-mpi.org/community/help/ |
this ticket started with Open MPI 5.0.5 and now it is 4.1.6. |
No sorry actually its not. I forgot to load the openmpi in the server thats why. The actual OpenMpi is 5.0.5. |
This mismatch between different versions of Open MPI is consistent with the error that started this issue. Please make sure your entire stack is compiled and run with a single version of Open MPI. |
I compiled the code Openmpi 5.0.5 version and also added "setNewSlipBoundary" and periodicity to the code: |
Dear OpenLB team,
I was running the Cylinder3d program while I am a new stl file which is extended at Y-axis and rest are the same. But while I am increasing the resolution (N) from 20 and above it does not run in ‘GPU’ but it is running well in ‘CPU’. When I used the ‘Gpu only’ config file it is giving an error: out of memory, so, I used the ‘Gpu openmpi’ config file and its giving error:
root@acmt-gpu:/home/achaudhury/olb-1.7r0/examples/laminar/cylinder3d# mpirun –allow-run-as-root -np 2 ./cylinder3d
./cylinder3d: symbol lookup error: /lib/x86_64-linux-gnu/libmpi_cxx.so.40: undefined symbol: ompi_mpi_errors_throw_exceptions
./cylinder3d: symbol lookup error: /lib/x86_64-linux-gnu/libmpi_cxx.so.40: undefined symbol: ompi_mpi_errors_throw_exceptions
————————————————————————–
prterun detected that one or more processes exited with non-zero status,
thus causing the job to be terminated. The first process to do so was:
Process name: [prterun-acmt-gpu-192644@1,0] Exit code: 127
————————————————————————–
root@acmt-gpu:/home/achaudhury/olb-1.7r0/examples/laminar/cylinder3d# mpicc -o cylinder3d cylinder3d.c
gcc: error: cylinder3d.c: No such file or directory
root@acmt-gpu:/home/achaudhury/olb-1.7r0/examples/laminar/cylinder3d# mpirun –version
mpirun (Open MPI) 5.0.5
Report bugs to https://www.open-mpi.org/community/help/
Please help me to solve this bug.
The text was updated successfully, but these errors were encountered: