Hans Pabst

@fstein93 I had no problems reusing the same recipe from the past for LIBXC 7.0, i.e., `configure` required no particular option(s) to enable the Fortran interface. Perhaps it's implicitly enabled...

I can confirm this too; it's the case since a long time. Btw, there are a number of components configured even if the respective component is turned off (OpenBLAS shall...

How about `git submodule update --init --recursive`?

GNU OpenMP and other OpenMP runtimes are warning about omp_set_nested being deprecated. The code suggests omp_set_max_active_levels is only available since OpenMP 5.0, which is not true.

It is at least unreliable as it fetches successfully sometimes...

The fix was merged two hours ago and there is no new release just yet. Also, the release cadence of DFTD4 does not suggest a fix "tomorrow". Is it possible...

I tested CPU_BIND=sockets since then. Though, it was suggested by CSCS Documentaiont for ALPS and CP2K in particular (https://docs.cscs.ch/software/sciapps/cp2k/#running-on-daint). I can change it back once the PR passes for the...

As a note, our Dashboard/CI (when based on cloud), can be non-reproducible or disruptive if we allow "equivalent or better SKUs" to be scheduled. Perhaps, we can also run more...

Ok, the COSMA thing I did not know (I mean perf. yes, but not about enabled vs not enabled before), thanks for collecting this!

Ok, "top 5 and rest" is fair enough, let me rephrase it for `diag_cu144_broy_6cpu_1gpu` which regressed from [189.323](https://dashboard.cp2k.org/archive/perf-cuda-volta/commit_8f495778c0b8e182213970d155e4a8f2a7e3c38a.txt) to [327.517](https://dashboard.cp2k.org/archive/perf-cuda-volta/commit_b31c023c7b0c70d05ce49c48e0a65b84b75e5dd8.txt) seconds. That's accompanied by "the 'rest' going through the roof",...