MFiX 26.1 is now available for download on the MFiX website.
This release includes a new flexible GSP (glued-sphere particle) model as well as numerous bugfixes addressing issues reported here.
The front-end has been ported from Qt5 to Qt6 GUI library. (Qt5 is no longer supported).
All users are encouraged to upgrade.
– Charles
Thanks for this.
A Q, you have bumped the required GLIBC version in the conda releases, is that by intention?
We’re running Alma 9.7, which was released in 2025, using glibc 2.34.
Hi @nicpa, welcome to MFiX
Alma 9.7 glibc 2.34 should not be a problem. Where do you see the bumped glibc version requirement?
Please paste the exact error messages you are seeing (avoid pasting the private download token) so we can diagnose exactly what is going wrong.
– Charles
Sorry, should have been more explicit.
I am referring to the conda installations, I followed the installation instructions, and got the following error when trying to run the solver step (I am not an MFix user, but a HPC supporter).
I did this:
conda activate mfix-26.1
mfix # this works just fine!
<choose RUN in the solver>
Then I got this error message:
h the following error:
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "/appl9/mfix/miniforge3/envs/XeonE5-2660v3-mfix-26.1/lib/python3.13/site-packages/mfixgui/pymfix.py", line 371, in <module>
main()
~~~~^^
File "/appl9/mfix/miniforge3/envs/XeonE5-2660v3-mfix-26.1/lib/python3.13/site-packages/mfixgui/pymfix.py", line 276, in main
M = ctypes.CDLL(so_path)
File "/appl9/mfix/miniforge3/envs/XeonE5-2660v3-mfix-26.1/lib/python3.13/ctypes/__init__.py", line 361, in __init__
self._handle = self._load_library(name, mode, handle, winmode)
~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/appl9/mfix/miniforge3/envs/XeonE5-2660v3-mfix-26.1/lib/python3.13/ctypes/__init__.py", line 401, in _load_library
return _dlopen(name, mode)
OSError: /lib64/libmvec.so.1: version `GLIBC_2.35' not found (required by /appl9/mfix/miniforge3/envs/XeonE5-2660v3-mfix-26.1/bin/mfixsolver.so)
So it’s your mfixsolver that got the problem.
Ok, that really should not be happening (the release should run on Alma 9), but if you are installing on HPC you will have better luck if you follow the Linux-HPC instructions. This install procedure has you skip installing the mfix-solver package and instead build it using your HPC’s compiler, MPI, etc.
Follow the instructions at https://mfix.netl.doe.gov/products/mfix/download/#hpc
You should delete the exisiting environment first:
conda env remove -n XeonE5-2660v3-mfix-26.1
Please let me know if this works for you. I will investigate the Alma failure.
– Charles
@nicpa The Conda packages are intended to be highly portable but due to an error they were linking against a sysroot with GLIBC 2.39. I regenerated the mfix-solver package (now mfix-solver-26.1-h94061f0_1.conda, the _1 being the incremented build number) to force a 2.28 sysroot. This should work on Alma 9 and pretty much any other OS out there newer than CentOS 6.
However, since you’re installing on HPC, I sill recommend the HPC install, where you link against the MPI that is already installed, instead of getting a possibly incompatible version from conda-forge.
Please let me know if this works for you.
– Charles
Thanks for the rebuild, we know it would justify installing with a good HPC installation, right now we are exploring it for our end-users, and tweaking small details, the MPI is compatible, so that isn’t a problem. But of course, there are performance to be gained.
Thanks for the quick response and help!
我的账号无法下载mfix软件，当我点击下载时会提示我让我注册，可是我已经注册并登陆了，当我填好注册信息时又提示我已经注册过
Registering is not sufficient to get access to the download page. You must wait for the registration to be approved. If the information you submitted is not valid, you will get an email saying the registration was rejected. It can take a few business days for a registration to be reviewed.