Skip to content
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

Make KokkosSparse::trsv wrap cuSPARSE's sparse triangular solve #48

Open
mhoemmen opened this issue Aug 21, 2017 · 4 comments
Open

Make KokkosSparse::trsv wrap cuSPARSE's sparse triangular solve #48

mhoemmen opened this issue Aug 21, 2017 · 4 comments
Assignees

Comments

@mhoemmen
Copy link
Contributor

mhoemmen commented Aug 21, 2017

Or any TPL. Alternately, get rid of KokkosSparse::trsv. The current implementation is sequential and implicitly assumes UVM.

mhoemmen pushed a commit to mhoemmen/Trilinos that referenced this issue Aug 21, 2017
@trilinos/tpetra @trilinos/ifpack2

Ifpack2's unit tests were failing for me in a debug build
(Kokkos_ENABLE_DEBUG:BOOL=ON), because Kokkos::DualView was
complaining about "concurrent modification" on device and host.  The
issue was that Tpetra::CrsMatrix::localSolve was sync'ing to device,
then running KokkosSparse::trsv.  The latter function currently runs
on host only, however.  This commit fixes the test failure by changing
localSolve, first to sync to host before the local solve, then to sync
back to device after the local solve.

General Tpetra preference is to move away from implementing things
like sparse triangular solve in Tpetra.  Ifpack2 has already started
supporting this.  See also the relevant kokkos-kernels issue,
kokkos/kokkos-kernels#48.
@srajama1
Copy link
Contributor

Deprecate and get rid of it. Calling it straight from ShyLU/HTS or LocalSparse containers such as in Ifpack2 is better option.

@crtrott
Copy link
Member

crtrott commented Sep 19, 2017

Ok we gonna do that and add MKL as well. The more complicated but better performing variants are gonna stay in a separate package.

@crtrott crtrott self-assigned this Sep 19, 2017
@mhoemmen
Copy link
Contributor Author

@crtrott Please feel free to move the current sequential sparse triangular solve to Tpetra if you want to remove it from KK. Just please put it in Tpetra::Details. Thanks!

@jhux2
Copy link

jhux2 commented Jun 6, 2023

@lucbv Is this issue still valid?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants