[Users] Status of support for multiple floating point precisions in Carpet?

[Erik Schnetter]

Erik

In principle both CCTK_REAL4 and CCTK_REAL8 should be supported.
CCTK_REAL16 never worked well because compilers disagree about its
layout and size.

(Integer types other than CCTK_INT4 should also be supported.)

To reduce compile time we then added the flags to disable support for
some of these types. It might be that this has bit-rotted and that
some newer code doesn't support CCTK_REAL4. It should be
"straightforward but tedious" to correct this.

I will have a brief look but might not have the time to correct all problems.

-erik

On Wed, Jan 17, 2024 at 9:41 AM Erik Keoni Wessel <ekwessel at arizona.edu> wrote:
>
> Hi,
> I'm currently working on a thorn development project where it would be helpful to be able to have a mix of single-precision grid functions and double-precision grid functions (a very unusual need, I know).
>
> When I had my thorn allocate a grid function of type CCTK_REAL4, I got a message upon running my configurations that I needed to enable support for this type in Carpet. After reading through the Carpet documentation and then digging deep into the source code for CarpetLib, I discovered that there seem to be flags to compile Carpet with support for different types. I passed -DCARPET_ALL_REAL in as part of CPPFLAGS in my optionlist, but attempting to compile with other real types leads to compile-time errors about undefined functions: apparently headers for prolongation functions for all the types are generated, but somehow the corresponding definitions aren't being be linked correctly.
>
> I'm not sure what is going wrong, but I am also unable to find any examples of how support for other types is supposed to be enabled, so I am unsure if I am doing this right.
>
> So my question is: what is the status of support for grid functions of types besides CCTK_REAL8 in Carpet? Does this feature work, and if so, how can I enable support for all Cactus-supported real types in Carpet?
>
> Cheers,
>
> — Erik Wessel
> Department of Physics
> University of Arizona
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users



-- 
Erik Schnetter <schnetter at gmail.com>
http://www.perimeterinstitute.ca/personal/eschnetter/