Add cuDSS support via mfem::CuDSSSolver#717
Conversation
cuDSS is a high-performance CUDA library for Direct Sparse Solvers. This commit enables the use of cuDSS in Palace by: - Wrapping mfem::CuDSSSolver in a new palace::CuDSSSolver; - Enabling the use of CuDSSSolver for KSP and WavePort; and - Adding cuDSS configuration options to CMake, for both Palace and MFEM.
The patch adding cuDSS to mfem has not yet been merged.
This patch has been merged into MFEM. The changes are already included in the version of MFEM required for cuDSS, and so the patch no longer applies cleanly.
|
I fixed the formatting, but I'm not sure what to do about the other failures. The license check is failing because the test assumes the header will match exactly, but I added a new copyright line to two files. One possible solution would be to rewrite the test to check for the copyright notice and SPDX license identifier separately, but I think that's outside the scope of this PR. I think the other tests are failing because they're using the Spack route of installation, which I didn't update. It's not obvious to me that there's a way to make this work, besides waiting for the mfem Spack package to expose cuDSS support. |
hughcars
left a comment
hughcars
left a comment
There was a problem hiding this comment.
We'll take this for a spin, one small comment on the patching process but generally looks good.
| set(EXTERN_MFEM_GIT_TAG | ||
| "d9d6526cc1749980a2ba1da16e2c1ca1e07d82ec" CACHE STRING | ||
| "e217864f168acb1d7bdb3fd7c5aa73ef81951237" CACHE STRING | ||
| "Git tag for external MFEM build" |
There was a problem hiding this comment.
For a POC this will work, but we have a facility for assembling patches for MFEM into palace. If we don't want to wait for the minor version bump in mfem (no idea when that'll be), the facility would be to assemble the patch for just the mfem PR, plucked onto our "palace head" state of mfem, which is 4.9 + the patch stack. This is a bit involved, so if you want I can do this for you (it's best achieved using spack nowadays).
There was a problem hiding this comment.
Yes, please, that would be very helpful. I only intended that commit to be temporary.
It probably still makes sense to wait for things to be merged into mfem as they are, before you assemble the patch -- I don't expect that the PR will change before it's merged, but I can't rule it out either.
hughcars
left a comment
hughcars
left a comment
There was a problem hiding this comment.
Hi @Pennycook,
I took a stab at cleaning things up as mentioned in the comments and there's a draft PR onto your fork at Pennycook#1. Most of that is noise from merging origin/main into your branch. If you merge into your fork it'll reduce down to a much smaller diff. The main thing was assembling the diffs.
2 participants
cuDSS is a high-performance CUDA library for Direct Sparse Solvers.
This commit enables the use of cuDSS in Palace by:
mfem::CuDSSSolverin a newpalace::CuDSSSolver;CuDSSSolverfor KSP and WavePort; andI've opened this as a draft for now because it depends on mfem/mfem#5124, which has not yet been merged.
An example of building with cuDSS: