[Llvm-bgq-discuss] bgclang r184137-20130617 on vesta/mira

[Erik Schnetter]

Hal

Thanks for the update! The new version compilers my code fine.

However, I am now encountering a wrong-code-generation issue. Essentially,
true && false evaluates to true, but of course in circumstances that are
more complex.

I have preprocessed source code that is 1.6 MB large
(/home/eschnett/src/vecmathlib-qpx2/test.i), but most code in there is
unused or commented out. The resulting file, when compiled with bgclang++11
-O3, runs very quickly and produces just a few lines of output. In fact, it
even runs on the front-end, producing the same error.

How should I continue to submit a bug report on this?

-erik



On Tue, Jun 18, 2013 at 3:00 PM, Hal Finkel <hfinkel at anl.gov> wrote:

> Hello again,
>
> I've upgraded the installation on vesta/mira to r184137-20130617. This
> incorporates a number of new upstream bug fixes and feature improvements,
> and also some QPX-specific bug fixes (as discussed recently on this list).
> I'll specifically note that this release picks up an upstream change which
> enables some limited loop autovectorization (QPX generation) at -O2
> (previously, there was no autovectorization of any kind until -O3).
>
> The archive will the patches, build and wrapper scripts has been posted:
> https://trac.alcf.anl.gov/projects/llvm-bgq/wiki
>
> I'd like to thank everyone again for their patience and feedback!
>
>  -Hal
>
> --
> Hal Finkel
> Assistant Computational Scientist
> Leadership Computing Facility
> Argonne National Laboratory
> _______________________________________________
> llvm-bgq-discuss mailing list
> llvm-bgq-discuss at lists.alcf.anl.gov
> https://lists.alcf.anl.gov/mailman/listinfo/llvm-bgq-discuss
>



-- 
Erik Schnetter <schnetter at cct.lsu.edu>
http://www.perimeterinstitute.ca/personal/eschnetter/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alcf.anl.gov/pipermail/llvm-bgq-discuss/attachments/20130618/fbd538d1/attachment.html>