Paul T. Bauman

This is tripping the ARefEE test. Today is a total horror show for me, but I'll try to get a unit going tomorrow and see if I can also replicate...

@bboutkov is cooking up a CPPUnit based test, but he mentioned that the patch as is still gave him a failure with mismatched n_local_dofs() and n_dofs() (he can chime in)....

I vote "yes" as well. I think if the lack of flexibility becomes as issue, I would think we could introduce a functor interface to allow the user to modify...

HI @artv3 Currently, you can't control the HIP compiler explicitly. The only knob here is `ROCM_PATH` which is what configure looks for for your ROCm installation (and defaults to `/opt/rocm`...

The primary HYPRE developers can comment, but you can build with HIP and without MPI. This requires the `--without-MPI` configure flag. On the other hand, when I build with MPI,...

So this issue is arising directly in HYPRE itself (with a forthcoming version of ROCm) in a way similar to as @dbarbier describes. hipcc -O2 -x hip -std=c++14 -I/home/pabauman/work/libs/rocprim/install/include -I/home/pabauman/work/libs/rocthrust/install/include...

Dup of #684?

No idea, but wanted to get the thought down before I forgot about it. "Design smell" means "too much design" or "needs better design than macro" or other?

IMHO, I'd say get rid of the Boost one, particularly since the C++11 one has been added. Then, if the user doesn't have a new enough compiler (anyone on RHEL/CentOS...

Fair enough. Was just hoping we could strip away Boost entirely and not have maintain it since it's been such a PITA.