This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: [RFC] Using SPEC CPU2006
- From: Kyrill Tkachov <kyrylo dot tkachov at arm dot com>
- To: Mikhail Maltsev <maltsevm at gmail dot com>, "gcc-help at gcc dot gnu dot org" <gcc-help at gcc dot gnu dot org>
- Cc: "H.J. Lu" <hjl dot tools at gmail dot com>, Vladimir Makarov <vmakarov at redhat dot com>
- Date: Fri, 05 Jun 2015 09:54:48 +0100
- Subject: Re: [RFC] Using SPEC CPU2006
- Authentication-results: sourceware.org; auth=none
- References: <55714E5D dot 1000105 at gmail dot com>
Hi Mikhail,
On 05/06/15 08:23, Mikhail Maltsev wrote:
Hi, all!
Could anyone please share some tips/recommendations w.r.t. testing
performance of code generated by GCC using SPEC CPU2006 benchmark (e.g.
recommended configs and patches for SPEC sources, if needed).
For example, 400.perlbench fails to link (GCC complains on multiple
definition of some string-related functions, like here:
https://vmakarov.fedorapeople.org/spec/spec2006.toolbox/gcc/log/20150117/747.log).
Is there a workaround?
For this particular issue, I believe it is because GCC 5 has moved to
a newer default C standard. Try adding -std=gnu89 to the
CPORTABILITY flags of perbench.
Kyrill
Also, I noticed that Vladimir stopped publishing SPEC 2006 results in
January 2015 (though current SPEC 2000 results are available). Is there
some policy on publishing benchmark results?