Jörg Saßmannshausen

Happy to comply. Does it help you to make the log-files of the tests available? As for hardware: GPFS mounted home directory, `IPv6` within the cluster, AMD Rome / Intel...

Do you build with `cmake` or `ninja` ? I am building with `ninja` and do the configuration with `meson`. I can switch to `gcc-13.3.0` as well and see if that...

I just build it: same problem. The same testjobs failed. What I noticed, and I might have missed that in the log file, was that: ``` 81/165 tblite:unit / double-dictionary...

@vladislavivanistsev Thanks for testing that. Which version of `gfortran` are you using here? I am wondering if that might be the underlying problem, we are using a too-new version of...

> Hmm, looks like that was a lucky shot. The tests actually passed when using `-O1 -fno-tree-vectorize -mieee-fp -mno-recip`, but I can't reproduce this. > > So, it seems like...

@tiwoe There are still some minor issues with the EasyConfig file, like missing white spaces and a blank line at the end of the file. I would suggest to run...

@boegelbot please test @ generoso

It looks like `PyTorch` has some failing tests: ``` distributed/_tensor/test_device_mesh 1/1 failed! distributed/test_c10d_common 1/1 failed! distributed/test_c10d_gloo 1/1 failed! distributed/test_c10d_nccl 1/1 failed! distributed/test_dynamo_distributed 1/1 failed! distributed/test_inductor_collectives 1/1 failed! == 2024-03-08 22:52:11,719...

@tiwoe You might want to sync your PR with development, something like this: ``` eb --sync-pr-with-develop NUMBER_OF_PR ``` Also, there is `PyTorch-2.1.2-foss-2023a-CUDA-12.1.1.eb` already merged in `EasyBuild-5.1.0`

@foxtran Thanks again for your response, much appreciated. > See [#1120](https://github.com/grimme-lab/xtb/issues/1120) In the EasyBuild community we are usually using releases, not commits, which is the reason why I was asking...