Loading a .msh mesh of big size into parallel::distributed::triangulation using p4est

[Phạm Ngọc Kiên]

Hi colleagues,

I have a problem with loading mesh of size more than 100,000 cells using MPI

I am writing to send you my test for loading grid with msh file into a parallel::distributed::triangulation.

The time for loading the grid is extremely huge. It takes most of the running time of my code.

I have tried both on my PC and on our lab's cluster, and it took about 400 seconds for the msh file containing 131936 cells.

In case of more than 200,000 cells, I cannot load the mesh.

I also tested that if we load the mesh to Triangulation<dim> triangulation, the time is much less than that of using parallel::distributed::triangulation.

However, with Triangulation<dim> triangulation, I don't think that it is suitable for run a large scale problem.

Could you please help me to address this problem?

Thank you very much.

Best regards,

Kien.

[Praveen C]

Just an observation running your code with deal.II-9.1.1

In debug mode, it runs at 65 sec.

In release mode, I get segmentation error.

Best

praveen

$ make

Scanning dependencies of target step-40

[ 50%] Building CXX object CMakeFiles/step-40.dir/step-40.cc.o

[100%] Linking CXX executable step-40

[100%] Built target step-40

$ ./step-40 

+---------------------------------------------+------------+------------+

| Total wallclock time elapsed since start    |      64.6s |            |

|                                             |            |            |

| Section                         | no. calls |  wall time | % of total |

+---------------------------------+-----------+------------+------------+

| load grid                       |         1 |      64.6s |     1e+02% |

+---------------------------------+-----------+------------+------------+

+---------------------------------------------+------------+------------+

| Total wallclock time elapsed since start    |  6.34e-06s |            |

|                                             |            |            |

| Section                         | no. calls |  wall time | % of total |

+---------------------------------+-----------+------------+------------+

+---------------------------------+-----------+------------+------------+

$ make release

Scanning dependencies of target release

[100%] Switching CMAKE_BUILD_TYPE to Release

-- Autopilot invoked

-- Run   $ make info  to print a detailed help message

-- Configuring done

-- Generating done

-- Build files have been written to: /Users/praveen/Downloads/loading-test

***

*** Switched to Release mode. Now recompile with:  $ make

***

[100%] Built target release

$ make

Scanning dependencies of target step-40

[ 50%] Building CXX object CMakeFiles/step-40.dir/step-40.cc.o

[100%] Linking CXX executable step-40

[100%] Built target step-40

$ ./step-40 

Segmentation fault: 11

[Phạm Ngọc Kiên]

Hi praveen,

If you run with 1 process(rank = 0) with ./step-40, it's fine.

I want to run with MPI on multiple processors with mpirun -np 4 ./step-40. In this case the trouble comes.

Thank you very much.

Best

Kien

Vào Th 7, 19 thg 10, 2019 vào lúc 13:45 Praveen C <cpra...@gmail.com> đã viết:

--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dealii+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/C9C81EF1-0A8D-4D7D-8297-88E46465556E%40gmail.com.