New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Sacado header brings in Kokkos and breaks compilation of deal.II with CUDA #6423
Comments
@davydden To reply to your comment spack/spack#7957 (comment)
No, I don't know anything about
with
There are only a few places in the library where
Thanks, I would appreciate it if you were to open a PR for this! |
Note that dealii+cuda+sacado works fine. I have a code that is using this configuration without any problem. The bug only appears when trilinos is compiled with kokkos which is the case if you use tpetra. |
The bug is really in Kokkos and there is nothing we can do about it. I have open an issue. |
kokkos/kokkos#1607 |
Everything seems to compile when I disable |
No, we actually only use the Epetra stack. The problem is if someone decides to install Trilinos with both the Epetra and the Tpetra stack. For some reason, Kokkos gets pull in and the problem arises. |
This is now fixed upstream and there is nothing to be done here anymore. |
as discussed in spack/spack#7957
include/deal.II/base/numbers.h
includesSacado.hpp
when compiled with Trilinos. That, in turn, leads to a chain that ends with includingKokkos_Macros.hpp
and gives compilation error when deal.II is built with CUDA:Links to similar Kokkos errors are here.
As a summary, it's a combination of dealii+CUDA and trilinos+sacado+kokkos.
In the original issue it was suggested to try using
Sacado_No_Kokkos.hpp
.EDIT:
Sacado_No_Kokkos.hpp
still includeskokkos
headers and gives the same compiling error.The text was updated successfully, but these errors were encountered: