# Performance of parallel MUMPS solver

**URL:** <https://getfem.discourse.group/t/performance-of-parallel-mumps-solver/65>\
**Category:** General\
**Created:** [March 6, 2025, 12:58pm UTC](https://getfem.discourse.group/t/performance-of-parallel-mumps-solver/65 "2025-03-06T12:58:48Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Konstantinos.Poulios](https://avatars.discourse-cdn.com/v4/letter/k/ac91a4/32.png) [@Konstantinos.Poulios](https://getfem.discourse.group/u/Konstantinos.Poulios)\
**Post date:** [March 6, 2025, 12:58pm UTC](https://getfem.discourse.group/t/performance-of-parallel-mumps-solver/65/1 "2025-03-06T12:58:48Z")

</div>

I have done some benchmarking of different MUMPS versions available on debian

Comparison between `libmumps-5.7` and `libmumps-ptscotch-5.7` packages, configuring getfem build respectively with

```auto
LIBS='-lmumps_common -lpord' --enable-paralevel=2 '--with-mumps=-lsmumps -ldmumps -lcmumps -lzmumps' -disable-openmp

```

and

```auto
LIBS='-lmumps_common_ptscotch -lpord_ptscotch' --enable-paralevel=2 '--with-mumps=-lsmumps_ptscotch -ldmumps_ptscotch -lcmumps_ptscotch -lzmumps_ptscotch' -disable-openmp

```

The produced python interface libraries link to the following shared libraries (right side with ptscotch)

 ![image](https://global.discourse-cdn.com/free1/uploads/getfem/original/1X/bf985afea4965e030e84932a05904f5dce19ac96.png)

Running a benchmark model with approx 250000 dofs gives the following timings (right side with ptscotch is much faster)

 ![image](https://global.discourse-cdn.com/free1/uploads/getfem/original/1X/27e8669b62a820a6db8339e1759ada8227589de6.png)

---

<div class="post-metadata">

**Author:** ![aeart](https://avatars.discourse-cdn.com/v4/letter/a/3e96dc/32.png) [@aeart](https://getfem.discourse.group/u/aeart)\
**Post date:** [May 21, 2025, 4:43am UTC](https://getfem.discourse.group/t/performance-of-parallel-mumps-solver/65/2 "2025-05-21T04:43:53Z")

</div>

Does the version of Mumps without the Scotch library use Metis or some internal reordering algorithm?

---

<div class="post-metadata">

**Author:** ![Konstantinos.Poulios](https://avatars.discourse-cdn.com/v4/letter/k/ac91a4/32.png) [@Konstantinos.Poulios](https://getfem.discourse.group/u/Konstantinos.Poulios)\
**Post date:** [June 2, 2025, 7:42am UTC](https://getfem.discourse.group/t/performance-of-parallel-mumps-solver/65/3 "2025-06-02T07:42:39Z")

</div>

my post refers to testing of debian mumps packages. The exact build options for these packages. with alternative ordering algorithms can be found here:

> **[debian · master · Debian Science Team / mumps · GitLab](https://salsa.debian.org/science-team/mumps/-/tree/master/debian?ref_type=heads)**
>
> MUMPS : a parallel sparse direct solver

Still, at runtime, one can change blas/lapack version in debian with the `update-alternatives --config ...` command, which is another factor to take intro account when benchmarking on debian.