[Wien] lapw0 stuck/drained with seemingly no error message upon launching SCF

[Yichen Zhang]

Dear Peter,

Yes, that was the .machines file when I attempted to test lapw in parallel mode. When running in sequential mode, “mv .machines .machines_disabled” was used. I hope that disables a possibility of running openMP?

Yes, indeed I was monitoring in top. When running sequential, only one CPU is occupied averaging around 95%-almost 100% usage, while for parallel mode with omp_global=2, two processes each occupy 100% of a CPU. The single lapw0 process usually has less than 100M memory usage, and the machine (64 GB unified memory) on average has 11-16 GB memory unused, depending on what other applications were running. And the lapw0 process hangs on the single CPU (in sequential mode) or two CPUs (in openMP parallel mode) forever, as we discussed.

Best regards
Yichen