[Openmp-dev] Public Benchmarks?

Hahnfeld, Jonas via Openmp-dev openmp-dev at lists.llvm.org
Tue Aug 23 05:21:04 PDT 2016


So that’s somehow expected: The default value for –fopenmp got flipped to libomp in 3.8, you needed –fopenmp=libomp for 3.7 which isn’t used in this article.

 

Cheers,

Jonas

 

From: Andrey Bokhanko [mailto:andreybokhanko at gmail.com] 
Sent: Tuesday, August 23, 2016 2:17 PM
To: Alexey Bataev
Cc: Hahnfeld, Jonas; LLVM-OpenMP (openmp-dev at lists.llvm.org)
Subject: Re: [Openmp-dev] Public Benchmarks?

 

There are a bit fresher results for clang 3.8 (though no comparison with gcc): http://www.phoronix.com/scan.php?page=article <http://www.phoronix.com/scan.php?page=article&item=llvm-clang-363738&num=3> &item=llvm-clang-363738&num=3

 

Yours,

Andrey

 

 

On Tue, Aug 23, 2016 at 3:12 PM, Alexey Bataev via Openmp-dev <openmp-dev at lists.llvm.org> wrote:

Phoronix results are not quite correct. Seems to me they have some problems with compilation flags. Tried to investigate their results a few years ago and got completely different numbers. 

Best regards, 

Alexey Bataev


23 авг. 2016 г., в 7:54, Hahnfeld, Jonas via Openmp-dev <openmp-dev at lists.llvm.org> написал(а):

There was in article one year ago: http://www.phoronix.com/scan.php?page=article <http://www.phoronix.com/scan.php?page=article&item=clang-svn-openmp&num=2> &item=clang-svn-openmp&num=2

I never came around to confirm the test results and it may look different for 3.9 now…

 

Regards,

Jonas

 

From: Openmp-dev [mailto:openmp-dev-bounces at lists.llvm.org] On Behalf Of Cownie, James H via Openmp-dev
Sent: Tuesday, August 23, 2016 1:48 PM
To: openmp-dev at lists.llvm.org
Subject: [Openmp-dev] Public Benchmarks?

 

I just noticed that the LLVM page in Wikipedia (https://en.wikipedia.org/wiki/LLVM ) contains the statement 

 

For run-time performance of the compiled programs, GCC formerly outperformed LLVM by 10% on average. <https://en.wikipedia.org/wiki/LLVM#cite_note-18> [18] <https://en.wikipedia.org/wiki/LLVM#cite_note-19> [19] Newer results indicate that LLVM has now caught up with GCC in this area, and is now compiling binaries of approximately equal performance, except for programs using  <https://en.wikipedia.org/wiki/OpenMP> OpenMP. <https://en.wikipedia.org/wiki/LLVM#cite_note-20> [20]

 

I believe that is wrong (their evidential link is from 2012!), and our OpenMP performance is now at least comparable with, if not better than, that of GCC, but I don’t have any public benchmarks to prove it. 

 

So, does anyone have any pointers to public articles that show that, so that we can go in and fix the Wikipedia article?

(Phoronix’ latest at http://www.phoronix.com/scan.php?page=article <http://www.phoronix.com/scan.php?page=article&item=gcc-61-clang39&num=1> &item=gcc-61-clang39&num=1 doesn’t seem to have any OpenMP codes in it, so that’s no use).

 

Thanks

 

-- Jim

James Cownie <james.h.cownie at intel.com>
SSG/DPD/TCAR (Technical Computing, Analyzers and Runtimes)

Tel: +44 117 9071438 <tel:%2B44%20117%209071438> 

 

---------------------------------------------------------------------
Intel Corporation (UK) Limited
Registered No. 1134945 (England)
Registered Office: Pipers Way, Swindon SN3 1RJ
VAT No: 860 2173 47

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

_______________________________________________
Openmp-dev mailing list
Openmp-dev at lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/openmp-dev


_______________________________________________
Openmp-dev mailing list
Openmp-dev at lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/openmp-dev

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/openmp-dev/attachments/20160823/36d19049/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5868 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/openmp-dev/attachments/20160823/36d19049/attachment.bin>


More information about the Openmp-dev mailing list