# Question about parmmg

**URL:** <https://community.freefem.org/t/question-about-parmmg/4139>\
**Category:** General Discussion\
**Created:** [November 20, 2025, 4:03pm UTC](https://community.freefem.org/t/question-about-parmmg/4139 "2025-11-20T16:03:40Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![zxss](https://avatars.discourse-cdn.com/v4/letter/z/bc8723/32.png) [@zxss](https://community.freefem.org/u/zxss)\
**Post date:** [November 20, 2025, 4:03pm UTC](https://community.freefem.org/t/question-about-parmmg/4139/1 "2025-11-20T16:03:40Z")

</div>

Dear all,

I implemented the 3D L shape Possion equation with parmmg for adaptation after watching this tutorial [https://www.youtube.com/watch?v=W08n2FqyFdc&t=9882s](https://www.youtube.com/watch?v=W08n2FqyFdc&t=9882s).

I implemented parmmg with both global mesh and distributed local meshes. The corresponding codes are as following:

[3Dparallel\_Parmmg\_global\_Lshape\_Poisson.edp](https://community.freefem.org/uploads/short-url/bh7ol3uxfjzcU5vdwAOFYbZkgtG.edp) (2.4 KB)

[3Dparallel\_Parmmg\_distributed\_Lshape\_Poisson.edp](https://community.freefem.org/uploads/short-url/qv8GQTLuZGUUMuhf0evGv0pF2dw.edp) (2.1 KB)

When I print the number of tetrahedrals of each process, I find that for distributed version, they vary a lot. For example, I use 12 prcocesses, at iteration 8, for global version, each processor contains almost the same number of tetrahedrals with the smallest 201032, largest 219798.

 ![Screenshot from 2025-11-20 15-55-12](https://canada1.discourse-cdn.com/flex030/uploads/freefem/original/2X/2/2d6e1ba8bfe14171a0c1e02989cb0f505fdf75c9.png)

while for distributed version, the smallest number is 198953, largest 243480. It varies a lot.

 ![Screenshot from 2025-11-20 15-57-44](https://canada1.discourse-cdn.com/flex030/uploads/freefem/original/2X/4/41e39e3eedc8fdd313f0e731217b8fa1c01a7263.png)

My question is, in the distributed version, is there a way to make each processor contain the same number of tetrahedrons as possible, since the processor containing the most tetrahedrons may reduce parallel performance?