Segmentation Fault (Core dumped) Error

Hi,

I am using the demo_stokes-iterative.py, with different meshes generated in Gmsh, but when I use a mesh with more degrees of freedom (around 500 000), gives the error :

Solving linear system of size 3782564 x 3782564 (PETSc Krylov solver).
Out of memory trying to allocate 1037075820 bytes
Segmentation fault (core dumped)

The computer has 8 CPU's Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz and 16G of RAM.

Do you know why this occurs?

asked Apr 25, 2016 by Ivelho FEniCS Novice (200 points)

1 Answer

You do not have enough memory.
You may use the command ´top´ to
monitor the memory usage.

answered Apr 25, 2016 by Kent-Andre Mardal FEniCS Expert (14,380 points)

Hi,

Thank you. I've confirmed with sar and top output that running the demo consumes all the 16G of RAM plus 16G of SWAP at the moment of the segmentation fault.

Can you give me, please, some explanation about the profile of memory consumption of fenics with stokes or navier-stokes problems?

Should I assume that this is the normal behaviour, and I have to run the code in a computational cluster to obtain results? Is this normal with a mesh with so few degrees of freedom?

commented Apr 26, 2016 by Ivelho FEniCS Novice (200 points)

...