Max Katz
Max Katz
Acknowledged, I was able to reproduce this on Perlmutter with Hypre v2.25.0 and the CUDA 11.7 from NVHPC.
Can you try this again on Perlmutter? I suspect this was fixed by https://github.com/AMReX-Codes/amrex/pull/2933
This can be done with #2147 as a first step, followed by a subsequent change to remove references to the `fluxes` MultiFab in post_timestep (which, I think, ends up being...
Yes, this is known behavior. See the last paragraph in this section of the documentation: https://amrex-codes.github.io/amrex/docs_html/Basics.html#initialize-and-finalize
Another way to do it would just be to assert if we're calling something in Arena after we have finalized (which could be checked with a static variable).
Using recent versions of CUDA I am able to compile this file in 1-2 minutes. I wonder if you are building on a platform where memory or CPU resources are...
This should only affect the situation where you would have OpenMP enabled at the compiler level but without using AMREX_USE_OMP, right? I am not sure I understand the motivation for...
> we also need to be able to set the stream priority on each GPU stream (in order to ensure the buffer packing kernel runs before the inner update kernels,...
I was thinking about this when I [recently added](https://github.com/AMReX-Codes/amrex/pull/1505) an option without documenting it. We could, if we wanted, take this as an opportunity to implement the runtime inputs system...
Castro doesn't need KLU but we do need Hypre for the radiation code, and I have struggled with this in the past. I too would like it to be the...