[llvm-dev] About OpenMP dialect in MLIR

[Nicolai Hähnle via llvm-dev]

On Thu, Feb 13, 2020 at 7:19 PM Johannes Doerfert via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> > 3. There would also be new sets of loop structures in new dialects when
> > C/C++ is compiled to MLIR. It would complicate the number of possible
> > combinations inside the OpenMP region.
>
> Is anyone working on this? If so, what is the timeline? I personally was
> not expecting Clang to switch over to MLIR any time soon but I am happy
> if someone wants to correct me on this. I mention this only because it
> interacts with the arguments I will make below.

I'd point out that from an engineering perspective, instead of
switching everybody from LLVM IR to MLIR, considering the relative
maturity of the projects the smarter move would be to switch folks
from MLIR to LLVM IR, by making LLVM IR itself extensible via
dialects.

It's understandable that MLIR was developed as a completely separate
thing, given that there was no proven example of an extensible IR.
However, now that such an example exists, integrating porting the
extensibility features that have been proven to work into LLVM IR is a
logical next step.

Cheers,
Nicolai
-- 
Lerne, wie die Welt wirklich ist,
aber vergiss niemals, wie sie sein sollte.