Alfio Lazzaro

> Sorry for the late reply. Yes, the problem is still there. However, the message seems to show up after the job finished so it doesn't bother too much for...

> I see the problem, although I still think that cp2k+cosma+nccl could outperform cp2k+cosma, at least on multi-gpu per node architectures, especially for gemm-dominant simulations. > > Another option is...

Added a banner in the wiki, see https://www.cp2k.org/howto:compile_on_windows_with_cygwin

I'm testing on my side the dashboard on AMD hardware (once per week?). I think the best solution is to go for the solution 1. and set the tests on...

Actually, I had a plan to install a Valgrind test on swan too (at least a SSMP version). I've started, but then I went to use valigrind4hpc and found a...

> > CP2K and DBCSR are still in a co-development state > > In other words: DBCSR has not yet acquired another big user besides CP2K. Given that this was...

Update the issue: DBCSR doesn't support the Makefile for the "general" public, the Makefile is hidden and made available only for CP2K building

By any chance, could you try to increase the stack size? ``` ulimit -s 256000 export OMP_STACKSIZE=256M ```

So, the conclusion is: we cannot use OpenMP and if we don't run on the GPU (because 2^n), the overall execution will be slower. Is this correct? If so, I...

> > 3.) multi-GPU and multi-node runs > ELPA (both 1 and 2) can run with several GPUs/node and several nodes (tested up to 256 nodes a 4 GPUs). Have...