From: Jeff Law <law@redhat.com>
To: Wilco Dijkstra <Wilco.Dijkstra@arm.com>,
"mjambor@suse.cz" <mjambor@suse.cz>
Cc: "gcc@gcc.gnu.org" <gcc@gcc.gnu.org>, nd <nd@arm.com>,
Steve Ellcey <sellcey@cavium.com>,
Richard Biener <richard.guenther@gmail.com>,
"pmenzel+gcc.gnu.org@molgen.mpg.de"
<pmenzel+gcc.gnu.org@molgen.mpg.de>
Subject: Re: How to get GCC on par with ICC?
Date: Fri, 15 Jun 2018 17:03:00 -0000 [thread overview]
Message-ID: <dcc58008-4c66-c8f5-e427-d22be0f1ef64@redhat.com> (raw)
In-Reply-To: <DB5PR08MB10302D16370CB306B77A6104837C0@DB5PR08MB1030.eurprd08.prod.outlook.com>
On 06/15/2018 05:39 AM, Wilco Dijkstra wrote:
> Martin wrote:
>
>> Keep in mind that when discussing FP benchmarks, the used math library
>> can be (almost) as important as the compiler. In the case of 481.wrf,
>> we found that the GCC 8 + glibc 2.26 (so the "out-of-the box" GNU)
>> performance is about 70% of ICC's. When we just linked against AMD's
>> libm, we got to 83%. When we instructed GCC to generate calls to Intel's
>> SVML library and linked against it, we got to 91%. Using both SVML and
>> AMD's libm, we achieved 93%.
>>
>> That means that there likely still is 7% to be gained from more clever
>> optimizations in GCC but the real problem is in GNU libm. And 481.wrf
>> is perhaps the most extreme example but definitely not the only one.
>
> You really should retry with GLIBC 2.27 since several key math functions were
> rewritten from scratch by Szabolcs Nagy (all in generic C code), resulting in huge
> performance gains on all targets (eg. wrf improved over 50%).
>
> I fixed several double precision functions in current GLIBC to avoid extremely bad
> performance which had been complained about for years. There are more math
> functions on the way, so the GNU libm will not only catch up, but become the fastest
> math library available.
And resolution on -fno-math-errno as the default. Setting errno can be
more expensive than people realize.
Jeff
next prev parent reply other threads:[~2018-06-15 15:26 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-15 11:48 Wilco Dijkstra
2018-06-15 17:03 ` Jeff Law [this message]
2018-06-15 18:01 ` Joseph Myers
-- strict thread matches above, loose matches on Subject: below --
2018-06-06 15:57 Paul Menzel
2018-06-06 16:14 ` Joel Sherrill
2018-06-06 16:20 ` Paul Menzel
2018-06-20 22:42 ` NightStrike
2018-06-21 9:20 ` Richard Biener
2018-06-22 0:48 ` Steve Ellcey
2018-06-06 16:22 ` Bin.Cheng
2018-06-06 18:31 ` Dmitry Mikushin
2018-06-06 21:10 ` Ryan Burn
2018-06-07 10:02 ` Richard Biener
2018-06-06 22:43 ` Zan Lynx
2018-06-07 9:54 ` Richard Biener
2018-06-07 10:06 ` Richard Biener
2018-06-08 22:08 ` Steve Ellcey
2018-06-09 15:32 ` Marc Glisse
2018-06-11 14:50 ` Martin Jambor
2018-06-22 22:41 ` Szabolcs Nagy
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=dcc58008-4c66-c8f5-e427-d22be0f1ef64@redhat.com \
--to=law@redhat.com \
--cc=Wilco.Dijkstra@arm.com \
--cc=gcc@gcc.gnu.org \
--cc=mjambor@suse.cz \
--cc=nd@arm.com \
--cc=pmenzel+gcc.gnu.org@molgen.mpg.de \
--cc=richard.guenther@gmail.com \
--cc=sellcey@cavium.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).