* [Bug rtl-optimization/68664] [6/7 Regression] Speculative sqrt in c-ray main loop causes large slow down
[not found] <bug-68664-4@http.gcc.gnu.org/bugzilla/>
@ 2017-02-06 21:15 ` hubicka at ucw dot cz
0 siblings, 0 replies; only message in thread
From: hubicka at ucw dot cz @ 2017-02-06 21:15 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68664
--- Comment #20 from Jan Hubicka <hubicka at ucw dot cz> ---
> Scheduling should never move very expensive instructions to places they
> are executed more frequently. This patch fixes that, reducing the
> execution time of c-ray by over 40% (I tested on a BE Power7 system).
>
> This introduces a new target hook sched.can_speculate_insn which returns
> whether the scheduler is allowed to speculate a given instruction. The
> rs6000 implementation disallows all divide and square root instructions.
>
>
> PR rtl-optimization/68664
> * target.def (can_speculate_insn): New hook.
> * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
> * doc/tm.texi: Regenerate.
> * sched-rgn.c (can_schedule_ready_p): Use the new hook.
> * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
> (rs6000_sched_can_speculate_insn): New function.
There was also regression on cray for x86-64
https://gcc.opensuse.org/c++bench-czerny/c-ray/
Is it the same issue?
Also with profile feedback perhaps you have enough info to tell that the
speculative path is almost as likely as the original placement.
I wonder if we can provie resonable default by RTX cost...
Honza
>From gcc-bugs-return-553177-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 21:24:32 2017
Return-Path: <gcc-bugs-return-553177-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 115228 invoked by alias); 6 Feb 2017 21:24:04 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 114816 invoked by uid 48); 6 Feb 2017 21:23:52 -0000
From: "hubicka at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/68664] [6/7 Regression] Speculative sqrt in c-ray main loop causes large slow down
Date: Mon, 06 Feb 2017 21:24:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 6.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: hubicka at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: segher at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 6.4
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: cc
Message-ID: <bug-68664-4-e2JfS75n6f@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-68664-4@http.gcc.gnu.org/bugzilla/>
References: <bug-68664-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00709.txt.bz2
Content-length: 438
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68664
Jan Hubicka <hubicka at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |hubicka at gcc dot gnu.org
--- Comment #21 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
The two regressions happened at 2017 01-13 and 2016 08-11
>From gcc-bugs-return-553178-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 21:32:18 2017
Return-Path: <gcc-bugs-return-553178-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 76437 invoked by alias); 6 Feb 2017 21:32:18 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 76218 invoked by uid 48); 6 Feb 2017 21:32:06 -0000
From: "segher at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/68664] [6/7 Regression] Speculative sqrt in c-ray main loop causes large slow down
Date: Mon, 06 Feb 2017 21:32:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 6.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: segher at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: segher at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 6.4
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-68664-4-CxiuKZKmIs@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-68664-4@http.gcc.gnu.org/bugzilla/>
References: <bug-68664-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00710.txt.bz2
Content-length: 694
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68664
--- Comment #22 from Segher Boessenkool <segher at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #20)
> There was also regression on cray for x86-64
> https://gcc.opensuse.org/c++bench-czerny/c-ray/
> Is it the same issue?
I don't think so. But I don't know much about that bug, it is something
with AVX I think? If you are talking about PR79224.
> Also with profile feedback perhaps you have enough info to tell that the
> speculative path is almost as likely as the original placement.
Maybe. The gain will be tiny though, afaics.
> I wonder if we can provie resonable default by RTX cost...
Not for rs6000 at least.
>From gcc-bugs-return-553179-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 21:59:40 2017
Return-Path: <gcc-bugs-return-553179-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 63945 invoked by alias); 6 Feb 2017 21:59:40 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 63901 invoked by uid 48); 6 Feb 2017 21:59:35 -0000
From: "cantabile.desu at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug libstdc++/71107] wstring_convert::from_bytes produces wide chars with the wrong byte order
Date: Mon, 06 Feb 2017 21:59:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: libstdc++
X-Bugzilla-Version: 6.1.1
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: cantabile.desu at gmail dot com
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-71107-4-YGzL5lH5Ee@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-71107-4@http.gcc.gnu.org/bugzilla/>
References: <bug-71107-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00711.txt.bz2
Content-length: 152
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71107
--- Comment #9 from Bitterblue <cantabile.desu at gmail dot com> ---
Still broken in GCC 6.3.1...
>From gcc-bugs-return-553180-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 22:14:57 2017
Return-Path: <gcc-bugs-return-553180-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 84387 invoked by alias); 6 Feb 2017 22:14:57 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 84247 invoked by uid 48); 6 Feb 2017 22:14:44 -0000
From: "roland.illig at gmx dot de" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug translation/79397] New: AltiVec spelled incorrectly in rs6000.opt
Date: Mon, 06 Feb 2017 22:14:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: translation
X-Bugzilla-Version: 7.0.1
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: roland.illig at gmx dot de
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: bug_id short_desc product version bug_status bug_severity priority component assigned_to reporter target_milestone
Message-ID: <bug-79397-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00712.txt.bz2
Content-length: 575
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79397
Bug ID: 79397
Summary: AltiVec spelled incorrectly in rs6000.opt
Product: gcc
Version: 7.0.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: translation
Assignee: unassigned at gcc dot gnu.org
Reporter: roland.illig at gmx dot de
Target Milestone: ---
In config/rs6000/rs6000.opt, the word AltiVec is sometimes spelled "Altivec"
and sometimes "AltiVec". The latter form should be used consistently.
>From gcc-bugs-return-553181-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 22:28:52 2017
Return-Path: <gcc-bugs-return-553181-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 111732 invoked by alias); 6 Feb 2017 22:28:52 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 111507 invoked by uid 48); 6 Feb 2017 22:28:39 -0000
From: "msebor at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/79398] New: misleading error static constexpr member function called in a constant expression before its definition is complete
Date: Mon, 06 Feb 2017 22:28:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: 7.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: msebor at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: bug_id short_desc product version bug_status bug_severity priority component assigned_to reporter target_milestone
Message-ID: <bug-79398-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00713.txt.bz2
Content-length: 1535
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79398
Bug ID: 79398
Summary: misleading error static constexpr member function
called in a constant expression before its definition
is complete
Product: gcc
Version: 7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: msebor at gcc dot gnu.org
Target Milestone: ---
The error below is a bit confusing: the definition of B::bar() is complete when
it's called.
I think I understand that the reason for the error below is actually that the
function is called before the definition of the class of which it's a member is
complete. The error should make that clear, although it seems that accepting
it (e.g., as an extension) would make static constexpr member functions quite a
bit more useful.
$ cat y.C && gcc -S -Wall -Wextra -Wpedantic y.C
struct A {
static constexpr int foo () { return 1; }
};
struct B: A {
static constexpr int bar () { return 2; }
enum E { e = B::foo () };
enum F { f = B::bar () };
};
y.C:8:23: error: ‘static constexpr int B::bar()’ called in a constant
expression before its definition is complete
enum F { f = B::bar () };
~~~~~~~^~
y.C:8:24: error: enumerator value for ‘f’ is not an integer constant
enum F { f = B::bar () };
^
>From gcc-bugs-return-553182-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 22:29:15 2017
Return-Path: <gcc-bugs-return-553182-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 112661 invoked by alias); 6 Feb 2017 22:29:15 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 112490 invoked by uid 48); 6 Feb 2017 22:29:02 -0000
From: "msebor at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/79398] misleading error static constexpr member function called in a constant expression before its definition is complete
Date: Mon, 06 Feb 2017 22:29:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: 7.0
X-Bugzilla-Keywords: diagnostic
X-Bugzilla-Severity: minor
X-Bugzilla-Who: msebor at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: keywords bug_severity
Message-ID: <bug-79398-4-AUCRNvAjKT@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-79398-4@http.gcc.gnu.org/bugzilla/>
References: <bug-79398-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00714.txt.bz2
Content-length: 352
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79398
Martin Sebor <msebor at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |diagnostic
Severity|normal |minor
>From gcc-bugs-return-553184-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 23:16:58 2017
Return-Path: <gcc-bugs-return-553184-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 64488 invoked by alias); 6 Feb 2017 23:16:58 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 64272 invoked by uid 55); 6 Feb 2017 23:16:45 -0000
From: "hubicka at ucw dot cz" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/68664] [6/7 Regression] Speculative sqrt in c-ray main loop causes large slow down
Date: Mon, 06 Feb 2017 23:16:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 6.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: hubicka at ucw dot cz
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution:
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: segher at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 6.4
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-68664-4-nsXPtck6j0@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-68664-4@http.gcc.gnu.org/bugzilla/>
References: <bug-68664-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2017-02/txt/msg00716.txt.bz2
Content-length: 1029
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68664
--- Comment #23 from Jan Hubicka <hubicka at ucw dot cz> ---
>
> I don't think so. But I don't know much about that bug, it is something
> with AVX I think? If you are talking about PR79224.
I see, we have separate PR for that, good ;)
>
> > Also with profile feedback perhaps you have enough info to tell that the
> > speculative path is almost as likely as the original placement.
>
> Maybe. The gain will be tiny though, afaics.
Usually profile is useful while driving speculation. I would add an
param that prevents moving insn when bb->count or bb->frequency increases
by large amount (like 100 times). Speculatively executing insn that will
be ignored 99% of times is probably not paying back...
Without profile that will probably only trigger for paths leading to abort()
or similar cases, but with profiles that may hit quite common scenarios.
Honza
>
> > I wonder if we can provie resonable default by RTX cost...
>
> Not for rs6000 at least.
>From gcc-bugs-return-553183-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Feb 06 23:16:11 2017
Return-Path: <gcc-bugs-return-553183-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 62790 invoked by alias); 6 Feb 2017 23:16:10 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 62773 invoked by uid 89); 6 Feb 2017 23:16:09 -0000
Authentication-Results: sourceware.org; auth=none
X-Virus-Found: No
X-Spam-SWARE-Status: No, score=-1.0 required=5.0 tests=AWL,BAYES_00,KAM_LAZY_DOMAIN_SECURITY,RP_MATCHES_RCVD autolearn=no version=3.3.2 spammy=profiles, Hx-languages-length:1004, driving, gain
X-Spam-User: qpsmtpd, 2 recipients
X-HELO: atrey.karlin.mff.cuni.cz
Received: from atrey.karlin.mff.cuni.cz (HELO atrey.karlin.mff.cuni.cz) (195.113.26.193) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Mon, 06 Feb 2017 23:16:08 +0000
Received: by atrey.karlin.mff.cuni.cz (Postfix, from userid 4018) id 031F681569; Tue, 7 Feb 2017 00:16:04 +0100 (CET)
Date: Mon, 06 Feb 2017 23:16:00 -0000
From: Jan Hubicka <hubicka@ucw.cz>
To: "segher at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
Cc: gcc-bugs@gcc.gnu.org
Subject: Re: [Bug rtl-optimization/68664] [6/7 Regression] Speculative sqrt in c-ray main loop causes large slow down
Message-ID: <20170206231604.GB20549@atrey.karlin.mff.cuni.cz>
References: <bug-68664-4@http.gcc.gnu.org/bugzilla/> <bug-68664-4-CxiuKZKmIs@http.gcc.gnu.org/bugzilla/>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <bug-68664-4-CxiuKZKmIs@http.gcc.gnu.org/bugzilla/>
User-Agent: Mutt/1.5.21 (2010-09-15)
X-IsSubscribed: yes
X-SW-Source: 2017-02/txt/msg00715.txt.bz2
Content-length: 905
>
> I don't think so. But I don't know much about that bug, it is something
> with AVX I think? If you are talking about PR79224.
I see, we have separate PR for that, good ;)
>
> > Also with profile feedback perhaps you have enough info to tell that the
> > speculative path is almost as likely as the original placement.
>
> Maybe. The gain will be tiny though, afaics.
Usually profile is useful while driving speculation. I would add an
param that prevents moving insn when bb->count or bb->frequency increases
by large amount (like 100 times). Speculatively executing insn that will
be ignored 99% of times is probably not paying back...
Without profile that will probably only trigger for paths leading to abort()
or similar cases, but with profiles that may hit quite common scenarios.
Honza
>
> > I wonder if we can provie resonable default by RTX cost...
>
> Not for rs6000 at least.
^ permalink raw reply [flat|nested] only message in thread