Max Katz
Max Katz
RFE: extend `cuda::std::numeric_limits` to the CUDA `__half` type. This should be legal according to cppreference as an extension of numeric_limits to an implementation-defined type.
This can cause problems on multi-GPU systems when the user is not intending to use GPU 0. This was not done consistently throughout all the notebooks, either, so I doubt...
## PR summary The support for `amr.plot_max_level` and `amr.small_plot_max_level` is added in AMReX-Codes/amrex#2825. ## PR checklist - [x] test suite needs to be run on this PR - [ ]...
We should have logic in the code that detects if we are likely going to have poor GPU performance, and aborts the run if so. The simplest logic here is...
In some cases at sharp composition gradients, the artificial viscosity may result in a term that is larger than the local flux, and in particular can flip the sign of...
It's possible for gamma to go negative when tracing if the timestep is much larger than it should be. Since the sound speed goes as sqrt(gamma), this results in c...
## PR summary The current default values (4 for both; the conservative energy formulation) are now the only option. Closes #156 ## PR checklist - [x] test suite needs to...
This setup resembles the sort of advection across the grid seen in wdmerger. When running at the problem's default (low) resolution, the star heats to a maximum temperature of over...
A non-subcycling AMR version of our code (in the Strang build, not the simplified SDC build) would do the following (assumes two levels for simplicity, and ignores burning for now):...
## PR summary This update makes the mass_fluxes array hold the full timestep's worth of mass fluxes at the end of finalize_advance. (This subtlety only matters if there were multiple...