First time here? Checkout the FAQ!
x
0 votes
by (210 points)
Hi all,

I have a question on the speed of the software. I am running basic 2D simulations on a square of 5cm by 5cm, with the TT2 model, and for 10 ms of simulations it takes 50s with 4 cores. However, running the same code with 12 cores, only speeds up the simulation with 5s.

I have the following system:

Linux version 4.15.0-106-generic (buildd@lcy01-amd64-016) (gcc version 7.5.0 (Ubuntu 7.5.0-3ubuntu1~18.04)) #107-Ubuntu SMP

and openCARP was installed recently from the source code. Does anyone know how I can speed up the simulations?

Thanks a lot!

Best

Nele

2 Answers

0 votes
by (3.6k points)
Hi Nele,

The scalabitlity of an mpi process is detemine by the hardware that you have available. More process does not mean that the simulation will run faster. Depending on your setup and your output requierements you can gain some speed. For example, you can increase dt, increase the resolution of the mesh, you can also consider other solving schemes such as mass lumping.

Best,

Jorge
0 votes
by (8.0k points)
Hi!

First I would suggest to make sure that you compile openCARP with optimization turned on. You do this by disabling the DEBUG flag in the my_switches.def and then recompile with "make clean all".

Next, please tell us the number of vertex points that your geometry has. Based on that information I can then tell you if the execution speed is as expected or there is some issue that needs to be tracked down.

Best wishes,

Aurel
Welcome to openCARP Q&A. Ask questions and receive answers from other members of the community. For best support, please use appropriate TAGS!
architecture, carputils, documentation, experiments, installation-containers-packages, limpet, slimfem, website, governance
MathJax.Hub.Config({ tex2jax: { inlineMath: [ ['$','$'], ["\\(","\\)"] ], config: ["MMLorHTML.js"], jax: ["input/TeX"], processEscapes: true } }); MathJax.Hub.Config({ "HTML-CSS": { linebreaks: { automatic: true } } });
...