This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug libstdc++/58437] [4.7/4.8/4.9 Regression] Sorting value in reverse order is much slower compare to gcc44
- From: "jmbnyc at gmail dot com" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Tue, 17 Sep 2013 17:32:50 +0000
- Subject: [Bug libstdc++/58437] [4.7/4.8/4.9 Regression] Sorting value in reverse order is much slower compare to gcc44
- Auto-submitted: auto-generated
- References: <bug-58437-4 at http dot gcc dot gnu dot org/bugzilla/>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58437
--- Comment #12 from Jeffrey M. Birnbaum <jmbnyc at gmail dot com> ---
Tammy,
We tested gcc 4.7.2, 4.6.2 and 4.4.3/5 (the bug is not in either 4.4.3/5). I
have gcc 4.8.1 on my laptop but have not tried it yet. I confirmed the issue by
compiling my test (almost identical to the one you submitted but using 500M
elements) on 4.4.5 and then moving the executable over to a box with 4.7.2
installed. the native compiled program performed poorly compared to the one
compiled with 4.4.5 (this also ruled out chip issues, i.e. haswell vs
sandybridge).
I knew something was wrong when my own single threaded merge sort that produces
a gradient instead of sorting the data in place was outperforming the std::sort
using -D_GLIBCXX_PARALLEL, i.e. parallel sort of 500M entries.
Best,
/JMB