Alfio Lazzaro

> By the way, generally it's fine to use Ubuntu 20.04, but the downside is that you get a very recent libc, so if you decide to run this docker...

Forget my previous comment, I see HIP is using 18 (https://github.com/cp2k/dbcsr/tree/develop/tools/docker)

Well, 20.04 is fine, of course...

That's a good idea... I will include it in v2.0

Back to this issue, currently we reset the GPU before each multiplication, see https://github.com/cp2k/dbcsr/blob/f35f901e4460980aa06757294463a1e6308f8dc9/src/mm/dbcsr_mm.F#L429 Should we reset for errors before any GPU calls?

- [ ] Refactor MPI wraps (see for example https://github.com/dftbplus/mpifx) - [ ] Remove F77 interface - [ ] Use F08 interface - [ ] Investigate the possibility to add...

Leak sanitizer is now included (see https://github.com/cp2k/dbcsr/commit/833941727e5e9c677ac784e048d3a4671679f606 ). It requires to add a [suppression file](https://github.com/cp2k/dbcsr/blob/develop/tools/docker/lsan.supp). Need to exclude OpenMPI because of some internal leaks. Added flags `-fsanitize=undefined -fsanitize=address -fsanitize-recover=all` only...

Well, following the gfortran convention, .F90 requires preprocessing. Currently, we take .F, run fypp and produce .F90 (under obj). The .f90 is used for included files.

@dev-zero did a great job to copy the entire history of the files commits we are using. Therefore the first set of commits is: https://github.com/cp2k/dbcsr/commits/develop?after=1da47620ea8ee33d68c6753e4c6adb1053c7fa73+2799&branch=develop Therefore, I would say that...

For the moment there are two issues related to put_block: 1) it doesn't check the dimensions of the new block versus the new block 2) it crashes if I try...