This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] x86-64: Compile branred.c with -mprefer-vector-width=128 [BZ #24603]
- From: "H.J. Lu" <hjl dot tools at gmail dot com>
- To: Jeff Law <law at redhat dot com>
- Cc: Joseph Myers <joseph at codesourcery dot com>, Florian Weimer <fweimer at redhat dot com>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Mon, 1 Jul 2019 12:20:37 -0700
- Subject: Re: [PATCH] x86-64: Compile branred.c with -mprefer-vector-width=128 [BZ #24603]
- References: <CAMe9rOoP_293SM=sYpuqx5Yg9K1a-F9GtN6g0DDFjHTZM5OkcA@mail.gmail.com> <alpine.DEB.2.21.1906101545000.29610@digraph.polyomino.org.uk> <CAMe9rOoyj_b=zWkUgcSKLxPLEuBA0Y9-75v+xqg0yT4HDXy56Q@mail.gmail.com> <87sgsg3yax.fsf@oldenburg2.str.redhat.com> <CAMe9rOpJ9J2LNNKwxP0QYgVJA0Ep9TWA21s_GNqacFPpfFx8Pg@mail.gmail.com> <CAMe9rOq1LqNOP4=yr5DCA=HyK+3wBqdW+FtRGLY1pWfgoEYDHA@mail.gmail.com> <87k1ds3xo3.fsf@oldenburg2.str.redhat.com> <87a7ektgjp.fsf@oldenburg2.str.redhat.com> <104ead0a-623d-6fea-ef3e-55bc3dc83459@redhat.com> <CAMe9rOqvb69AkKs8Zo-XUAme=JWm+=Pdxk99p1wnDYB6kuUpTQ@mail.gmail.com> <6eff631b-c039-f726-18e8-28231d398e59@redhat.com> <87o92wc9he.fsf@oldenburg2.str.redhat.com> <CAMe9rOrLXTTLLmqkLwO6vQpHxx2DumiH+jM-VC6mxbG47GSEqQ@mail.gmail.com> <alpine.DEB.2.21.1906172137110.4463@digraph.polyomino.org.uk> <53114b6b-a1c5-4f99-7ec7-0cba2d4980e3@redhat.com> <CAMe9rOoWVvVF5Hd25ZvxEh0kHTMnsTiXpb808ByWY13g7J384w@mail.gmail.com>
On Thu, Jun 20, 2019 at 9:30 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Mon, Jun 17, 2019 at 2:41 PM Jeff Law <law@redhat.com> wrote:
> >
> > On 6/17/19 3:37 PM, Joseph Myers wrote:
> > > On Mon, 17 Jun 2019, H.J. Lu wrote:
> > >
> > >> On Mon, Jun 17, 2019 at 12:12 PM Florian Weimer <fweimer@redhat.com> wrote:
> > >>>
> > >>> * Jeff Law:
> > >>>
> > >>>> ISTM that 8, 9, 10 would use the new flag. 11 would issue an error
> > >>>> which would trigger a reinvestigation roughly a year from now.
> > >>>
> > >>> Do you suggest to put in a #warning for GCC 11, so that people can
> > >>> configure with --disable-werror and still build with GCC 11?
> > >>>
> > >>> Given that GCC 11 will probably add other warnings that break the build,
> > >>> this proposal isn't entirely unreasonable (even though I generally
> > >>> dislike such time bombs).
> > >>>
> > >>
> > >> Since there is no issue in source, this requires an artificial warning
> > >> purely based on GCC version. We can place GCC version check in
> > >> configure script with explicit -mprefer-vector-width=128 support.
> > >>
> > >> --enable-mprefer-vector-width:
> > >>
> > >> 1. Default. For GCC 8, 9, 10, use -mprefer-vector-width=128.
> > >> For GCC 11 and above, issue a configure error.
> > >
> > > I don't think anything causing the build to break with newer GCC versions
> > > is appropriate.
> > Fair enough -- hence a bit of a step away from the decision process in
> > my prior message on this subject. I threw it out as a way to address
> > the issue Florian raised, but I strongly believe it's up to the glibc
> > team to decide how they want to go forward.
> >
>
> I'd like to check in my patch to close the 40% performance gap with GCC
> 8, 9 and 10. We will monitor performance changes in the future GCC.
>
> Thanks.
>
I will check in my patch tomorrow.
Thanks.
--
H.J.