Hans Pabst
* Ensure BLAS and LAPACK are used as configured `cmake [...] -DLAPACK_LIBRARIES=... -DBLAS_LIBRARIES=.. [...]` - The CMake recipe may prefer BLAS and LAPACK if found in standard locations like `/usr`....
Looking at the CUDA Volta benchmarks like https://dashboard.cp2k.org/archive/perf-cuda-volta/index.html, I wonder if we should generally track and denote the total time to solution and potentially improve the graphing of results. If...
- TTS on Alps using CPU only (single node) to be improved. - The workload is best with square number for rank-count.
It seems `dbcsr_mm_multrec.F:309` can crash with OOB access if ranks have unequal number of threads? A test script looks like: ```bash #!/usr/bin/env bash NPROC=$(nproc) MYVAR=$1 VAR=$((MYVAR