Ben Prather

I'm a computational astrophysicist, currently a postdoc at LANL. I'm interested in black hole accretion problems and in making use of the Event Horizon Telescope results, but also the guts...

I can verify it's present in Parthenon proper when using the Kokkos `nvcc_wrapper` with `nvc++` to compile for device (incidentally the process also uses up to ~11GB of RAM by...

Just tried to compile with NVHPC/NVC++ 23.9 and... it segfaults. The crash is in cpp2 as you might expect, when processing `params.cudafe1.cpp`. So, if we can't figure out how to...

This issue has gotten worse (>30min compiles with NVHPC 22.11 and above, which may as well be a compile error). It doesn't disappear even if I place ``` #ifndef __CUDA_ARCH__...

I am abusing this, actually! How did you guess? The central problem this solves for me is I want a single simulation with two distinct phases: 1. Clean up magnetic...

Hmm, maybe `CopyFrom` is all I need. Certainly for splitting off the solver step, I should be using that instead of copying and removing. I think the issue is that...

You know, looking back I did "implement" it in my branch for KHARMA, and then just forgot. It's just one line, `varMap_.erase(label);` ...meaning it probably plays hell with memory, but...

Yep, definitely not worth reviewing now. I also have some changes & fixes still queued up for this, it's still evolving a little as I firm up how I'm using...

Linking issue #1046 so we remember it before people start using this stuff hard. Maybe it's already fixed in this PR?

Cool, I figured it might be useful. Using the coarse buffers is a good idea, I bet I can whip up the easy 1:2 version in a few weeks here...