IS this project not being actively maintained?

[L-A-F-B]

I looked at issues, and one person filed issue 74 in June. It is an issue I have seen. I added a comment with another failure pathway.

I would like to take one of your CUDA examples and put it through your toolchain and get out LLVM IR. Could you please demonstrate how you think this should work?

None of the CUDA tests pass, by the way, when I do the check after the build. I can provide more details if I think there is a point -- like someone actively monitoring these issues.

[whisperity]

Unfortunately, I don't think so. There's been very little activity, barely any progress on the post-incubator state, no new features, a backlog on the mailing list as well, and no publications either. Looking at the main authors, it seems they have moved on to other projects at @EnzymeAD. Seems Polygeist never had that "woofh" to ignite a proper take-off.

Alternatively, the MLIR C/C++ group around ClangIR (CIR, https://github.com/llvm/clangir, https://github.com/llvm/llvm-project/tree/main/clang/lib/CIR) is rather active, you might wish to read up on that (https://discourse.llvm.org/t/mlir-c-c-frontend-working-group-mon-sep-8th/88048) or participate! CIR has at least some interested developers currently behind it whose day jobs are paying for said interest. 🙂 Unfortunately, though, CIR is also plagued by it's two-facedness, namely that they have two pipelines: source code → CIR → backend (the direct codegen mode) and source code → CIR → other MLIR dialects → something(?) (the through-MLIR mode), and the latter one which might be more interesting for you (going from CIR to some SCF and GPU dialect hybrid) is the one that's having less traction in the upstream.

[wsmoses]

hi, sorry for the delay (was in hospital last night)!

well yes and no.

So polygeist is alive and well but the main dev work is not currently in this repo.

The current updated version [with all the passes/lowering/features/etc] is inside of this repo (https://github.com/EnzymeAD/Enzyme-JaX) and vague docs for here: https://docs.google.com/document/d/1wgSuKWtlaFkqKTi5EDc6PbKb88_OrSKf70XtqeFST3M/edit?tab=t.0

We really need to upstream the relevant passes to LLVM proper [and therefore all users, including integration with CIR] can benefit, as well as update this repo's polygeist passes to the latest version in Enzyme-JaX.

[peihunglin]

We have tried using Polygeist to extract affine computation patterns from scientific codes written in C++17. Our goal is pattern extraction for proofing/analysis, not optimization. In doing so, we’ve encountered several issues related to C++ feature support, and we’re wondering whether there are plans to further develop Polygeist’s support, here or under EnzymeAD, for modern C++.