From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 111771 invoked by alias); 15 Jan 2019 16:18:40 -0000 Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org Received: (qmail 111759 invoked by uid 89); 15 Jan 2019 16:18:39 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-10.4 required=5.0 tests=BAYES_00,GIT_PATCH_2,GIT_PATCH_3,KAM_ASCII_DIVIDERS,KAM_LOTSOFHASH,KAM_NUMSUBJECT,SPF_HELO_PASS,SPF_PASS,UNPARSEABLE_RELAY autolearn=ham version=3.3.2 spammy=Qing X-HELO: userp2120.oracle.com Received: from userp2120.oracle.com (HELO userp2120.oracle.com) (156.151.31.85) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Tue, 15 Jan 2019 16:18:37 +0000 Received: from pps.filterd (userp2120.oracle.com [127.0.0.1]) by userp2120.oracle.com (8.16.0.22/8.16.0.22) with SMTP id x0FG9QAe157186; Tue, 15 Jan 2019 16:18:34 GMT DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=oracle.com; h=content-type : mime-version : subject : from : in-reply-to : date : cc : content-transfer-encoding : message-id : references : to; s=corp-2018-07-02; bh=Kk+VV6I6RyoPxI1OTUZ6+uR0qN8g7ySwKm4bTUwH3ak=; b=glgYOcPiaqPlR9gaa71NYCz6SLxpncftK5nKDa3HXQ21jcFzKLpBFRbGbqlK+e/mio14 PN1w+jESLzLa/RnFrQMVexnjiunz3dOj1rWEfnayjOCiiyLLMn9RnGHbQ2x5qvbfk7nA h0tSpbdXV1C5k5FetEb9G4G1OayovLA/a669q4PtTmFPFb8CbT3PbIpOkR3NHOZ0YQ3M kZ66VeAsIfSRhseJp6OU/oOVe0B/h81X1HUXgu4B/WDz5nzLzS84+fH/lgH/hRfhFw5O 4Il/iA0KYl0u1CsCx5oT7gZLjIICXHYvmneH6yFTvvgxUdEiF3g6Qbb+fKK71h4oDbAJ cg== Received: from aserv0021.oracle.com (aserv0021.oracle.com [141.146.126.233]) by userp2120.oracle.com with ESMTP id 2pybjs4tw5-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=OK); Tue, 15 Jan 2019 16:18:34 +0000 Received: from aserv0121.oracle.com (aserv0121.oracle.com [141.146.126.235]) by aserv0021.oracle.com (8.14.4/8.14.4) with ESMTP id x0FGIXjx020381 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=OK); Tue, 15 Jan 2019 16:18:33 GMT Received: from abhmp0014.oracle.com (abhmp0014.oracle.com [141.146.116.20]) by aserv0121.oracle.com (8.14.4/8.13.8) with ESMTP id x0FGIX83015019; Tue, 15 Jan 2019 16:18:33 GMT Received: from dhcp-10-159-239-236.vpn.oracle.com (/10.159.239.236) by default (Oracle Beehive Gateway v4.0) with ESMTP ; Tue, 15 Jan 2019 08:18:33 -0800 Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 (Mac OS X Mail 10.3 \(3273\)) Subject: Re: Set inline-unit-growth to 40 From: Qing Zhao In-Reply-To: <20190115151442.ge5brvlz5myq3fx7@kam.mff.cuni.cz> Date: Tue, 15 Jan 2019 16:18:00 -0000 Cc: gcc-patches@gcc.gnu.org Content-Transfer-Encoding: quoted-printable Message-Id: <17442032-EE39-4DC4-BE50-5263FE5CB3FB@oracle.com> References: <20190112183228.cajeytm4r3d4xyjs@kam.mff.cuni.cz> <20190115151442.ge5brvlz5myq3fx7@kam.mff.cuni.cz> To: Jan Hubicka X-IsSubscribed: yes X-SW-Source: 2019-01/txt/msg00853.txt.bz2 Okay, I see. thanks. Qing > On Jan 15, 2019, at 9:14 AM, Jan Hubicka wrote: >=20 >> Hi, Honza, >>=20 >> in addition to the code size problems, there are several runtime regress= ion for the SPEC: (If I read the table correctly, if not, let me know) >>=20 >> SPEC/SPEC2006/INT/483.xalancbmk 146.131 4.89% >=20 > This test seems to be just a noise, if you look on the mainline plots > there is no noticeable regression and you can see differences +- 4% in > last 10 runs. >>=20 >> SPEC/SPEC2006/FP/436.cactusADM 130.967 8.07%=09 >>=20 >> SPEC/SPEC2017/INT/520.omnetpp_r 395.582 4.98%=09 >=20 > Here you can see in the graph both boxes to be yellow which means that > binary did not changed nor the size changed. It is just measurement error= it seems. >>=20 >> SPEC/SPEC2006/FP/435.gromacs 182.555 11.73%=09 >=20 > I plan to check on gromacs > This is LTO+PGO which will be rerun only in day or two so I will see if > the same regression stays on mainline. >>=20 >> SPEC/SPEC2017/INT/541.leela_r 452.333 4.17%=09 >=20 > This was already taken by daily testers and regression does not > reproduce, so it seems to be noise too. >=20 > Honza >>=20 >> do we have plan to study and fix these run-time regression? >=20 >>=20 >> thanks. >>=20 >> Qing >>=20 >>> On Jan 12, 2019, at 12:32 PM, Jan Hubicka wrote: >>>=20 >>> Hello, >>> this patch sets inline-unit-growth to 40. The performance changes are >>> - Firefox, LTO >>> https://treeherder.mozilla.org/perf.html#/compare?originalProject=3Dtry= &originalRevision=3Df7bd026e1a931b9a284d1c85c2577a72dd592820&newProject=3Dt= ry&newRevision=3D74889968abcc688b8d161863566ed273c0401ee4&framework=3D1&fil= ter=3Dopt&showOnlyComparable=3D1&showOnlyImportant=3D1 >>> After fixes to inlining priorities this makes difference without >>> profile feedback only. >>>=20 >>> Code size growth is about 9.15% with LTO and 3.95 with LTO and profile >>> feedback. >>> - Firefox noLTO >>> https://treeherder.mozilla.org/perf.html#/compare?originalProject=3Dtry= &originalRevision=3Dc902b72340a3dca3114f58578c1c8f3e6a1cd89c&newProject=3Dt= ry&newRevision=3D4974da6f92c144a9c09765b56a564a640069ddb9&framework=3D1&sho= wOnlyComparable=3D1&showOnlyImportant=3D1 >>> With about 7% code size growth >>> - SPEC >>> https://lnt.opensuse.org/db_default/v4/CPP/latest_runs_report?num_runs= =3D10&min_percentage_change=3D0.02&revisions=3D46e2bd1143b5c60af814916d7673= 879b34ceb3f6%2Cc0d79cfe9c4ec30823480f2f9b256600e8e3899f >>> - C++ benchmarks >>> https://lnt.opensuse.org/db_default/v4/SPEC/latest_runs_report?num_runs= =3D10&all_changes=3Don&min_percentage_change=3D0.02&revisions=3D46e2bd1143b= 5c60af814916d7673879b34ceb3f6%2Cc0d79cfe9c4ec30823480f2f9b256600e8e3899f >>>=20 >>> I am not entirely happy about the code-size/performance tradeoffs but it >>> is concerned only for programs built with -O3 or having too many inline >>> keywords. I have looked into inlining decisions for Firefox, HHVM and >>> Clang and inliner gets out of growt bounds way too early and some of >>> more performance aware projects already sets the limit up. >>>=20 >>> I will tune other metrics down to handle some of the code size problems. >>>=20 >>> Honza >>>=20 >>> Index: ChangeLog >>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >>> --- ChangeLog (revision 267882) >>> +++ ChangeLog (working copy) >>> @@ -1,3 +1,7 @@ >>> +2019-01-05 Jan Hubicka >>> + >>> + * params.def (inline-unit-growth): Set to 40. >>> + >>> 2019-01-12 Jakub Jelinek >>>=20 >>> * tree-ssa-loop-ivopts.c (find_inv_vars): Fix a comment typo. >>> Index: params.def >>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >>> --- params.def (revision 267882) >>> +++ params.def (working copy) >>> @@ -227,7 +227,7 @@ DEFPARAM(PARAM_LARGE_UNIT_INSNS, >>> DEFPARAM(PARAM_INLINE_UNIT_GROWTH, >>> "inline-unit-growth", >>> "How much can given compilation unit grow because of the inlining (in= percent).", >>> - 20, 0, 0) >>> + 40, 0, 0) >>> DEFPARAM(PARAM_IPCP_UNIT_GROWTH, >>> "ipcp-unit-growth", >>> "How much can given compilation unit grow because of the interprocedu= ral constant propagation (in percent).", >>=20