This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug bootstrap/68404] [6 Regression] PGO/LTO bootstrap failure on ppc64le
- From: "meissner at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Wed, 10 Feb 2016 18:12:25 +0000
- Subject: [Bug bootstrap/68404] [6 Regression] PGO/LTO bootstrap failure on ppc64le
- Auto-submitted: auto-generated
- References: <bug-68404-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68404
--- Comment #28 from Michael Meissner <meissner at gcc dot gnu.org> ---
I agree the 'fix' was a work around. I think the real fix is to make the
addresses into normal addresses but I need some way of marking the address to
say this is ok as a fused address, but not as a normal non-fused address.
However, such changes are not appropriate for stage4.
So I will come up with a patch to just disable -mcpu=power8 from turning fusion
on. I did some tests recently on a power8, and fusion as it is currently
implemented does not help the Spec 2006 benchmarks all that much (most
benchmarks are the same speed or 1% slower, with perlbench being 2% slower if I
disable fusion).