[llvm-dev] Profile-Guided Compiler Optimization with LLVM/Clang

[Anja Gerbes via llvm-dev]

Dear LLVM community,


I am very happy to become a member of this great community. At the German
Climate Research Center (DKRZ) I am a PhD student and write my thesis in
python. The general topic is Compiler Optimization in High Performance
Computing. My use cases are climate models, written in C/C++.
In Frankfurt I am working at a HPC Cluster Center and my duties are to
monitor and to optimize the usage of the cluster.

Last year in Saarbrücken was my first time to visit LLVM developers
meeting. There I presented my poster about "Intelligent selection of
compiler options to optimize compile time and performance" (
http://llvm.org/devmtg/2017-03//assets/slides/intelligent_se
lection_of_compiler_options_to_optimize_compile_time_and_performance.pdf)

I would like to learn from all of you and would like to start a conversion
about the specific topic: Profile-Guided Compiler Optimization with
LLVM/Clang.

We've already tested several optimization flags and want to continue to
understand and exploit the capabilities of compilers for a better
understanding: what performance a system can best achieve.

The AIMES project (https://wr.informatik.uni-ham
burg.de/research/projects/aimes/start) has already developed a number of
higher-level language extensions (DSL) to support climate modeling, as well
as the development of a source-to-source translation tool that translates
domain-specific code for various architectures (GPU, CPU).

My job will be, basically to optimize compilers that HPC applications can
run more efficiently. This provides an interface to the DSL. With the help
of DSL we can make transformations that a compiler may not make due to
language restrictions. The code should then be translated into a form that
can achieve the best possible performance from the compiler.

I really appreciate your reply.

Kind Regards

Anja
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20180507/7298219e/attachment.html>