Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.compilers > #3599 > unrolled thread

Fully integrating the Flang Fortran compiler with standard MLIR

Back to article view | Back to comp.compilers

Contents

  Fully integrating the Flang Fortran compiler with standard MLIR John R Levine <johnl@taugh.com> - 2024-09-30 18:10 +0100
    Re: Fully integrating the Flang Fortran compiler with standard MLIR "Steven G. Kargl" <sgk@REMOVEtroutmask.apl.washington.edu> - 2024-10-03 04:23 +0000

#3599 — Fully integrating the Flang Fortran compiler with standard MLIR

The authors modified Flang, the LLVM Fortran compiler, to use MLIR rather
than its private IR, and found that the code got better, but still not as
good as the Cray compiler.

https://arxiv.org/abs/2409.18824

Regards,
John Levine, johnl@taugh.com, Taughannock Networks, Trumansburg NY
Please consider the environment before reading this e-mail. https://jl.ly

[toc] | [next] | [standalone]

#3600

On Mon, 30 Sep 2024 18:10:06 +0100, John R Levine wrote:

> The authors modified Flang, the LLVM Fortran compiler, to use MLIR rather
> than its private IR, and found that the code got better, but still not as
> good as the Cray compiler.
>

The paper, like many involving comparisons/benchmarks,
is IMHO flawed. The author does not give the options
used with the various compilers.  The author also used
a 3 year old version of gfortran from 2021 and compares
it to 2023 versions of flang.  I cannot find the release
date for Cray Fortran 15.0.0.

PS: The Fortran compiler available from LLVM github is
named flang-new.  There is no compiler named flang.

--
steve
[In case it wasn't clear, I'm posting links to these draft papers because
they look interesting, not necessarily because I think they are brilliant
and insightful.  Asssessing those is for all you guys. -John]

[toc] | [prev] | [standalone]

Back to top | Article view | comp.compilers

csiph-web