FLANG and performance analysis

[Jan Lehr]

Hello flang-dev list,
I am a student research assistant from TU Darmstadt University, Germany, and attended the FORTRAN workshop at EURO LLVM in London.
During the discussion the topic of "performance analysis" and how to leverage the compiler surfaced, if I recall correctly.
Since my supervisor is working on that particular topic, I am a little involved in that topic as well.
His research is about a generic direct measurement abstraction (as opposed to sampling) which enables users to precisely specify which parts of a program are of interest for performance measurement and how a compiler should provide instrumentation/measurement of those constructs.
We have a prototypic implementation on a source-to-source basis for C/C++ using the ROSE compiler infrastructure and I recently ported parts of this abstraction layer and key concepts to Clangs tooling library as well as LLVM.
I am curios to what extend the community would be interested in such an abstraction and whether and how we could contribute.
Best regards,
JP

[John Leidel]

Jan, thanks for attending the talk in London.  I believe that performance analysis and performance tooling will be an important topic of discussion as we move the project forward.  IMHO, Fortran users tend to be rather performance conscious given the historic use of the language in scientific/numerical codes.  

Do you have any papers/documentation/etc regarding your work to port the concepts to the LLVM/Clang tooling infrastructure? I believe this would be a good way to present your concepts to the Flang/Fortran community, thus garnering support for an official feature request.  

cheers

john 

--
You received this message because you are subscribed to the Google Groups "flang-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flang-dev+...@googlegroups.com.
To post to this group, send email to flan...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/flang-dev/1832eda2-2b66-4133-a659-cf0413e841b3%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jan Lehr]

Thanks for agreeing that performance tooling is probably going to be an important topic in the future and sorry for the late response.
The assignment isn't graded yet and I will have to wait until this process is done. After this is done it should be available.
While this work follows a source-to-source approach, we are also playing around with integrating this in Clang (codegen) itself. However, I am not completely sure about the time this work will be available.

Cheers
Jan

[Damian Rouson]

On Wednesday, April 29, 2015 at 12:07:51 AM UTC-7, Jan Lehr wrote:

Hello flang-dev list,
I am a student research assistant from TU Darmstadt University, Germany, and attended the FORTRAN workshop at EURO LLVM in London.

 It's always exciting to see students interested working on Fortran.  

Please spell the language "Fortran".  The mixed-case spelling has been in the standard since at least Fortran 90 I believe. 

I won't comment on the remainder of the email as I'm not a FLANG developer -- just monitoring the list to keep up on the latest developments in Fortran compilers.

Damian