From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 22187 invoked by alias); 3 Aug 2011 21:46:03 -0000 Received: (qmail 22168 invoked by uid 22791); 3 Aug 2011 21:46:02 -0000 X-SWARE-Spam-Status: No, hits=-2.6 required=5.0 tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,RP_MATCHES_RCVD,SPF_HELO_PASS X-Spam-Check-By: sourceware.org Received: from smtp-out.google.com (HELO smtp-out.google.com) (74.125.121.67) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Wed, 03 Aug 2011 21:45:46 +0000 Received: from hpaq6.eem.corp.google.com (hpaq6.eem.corp.google.com [172.25.149.6]) by smtp-out.google.com with ESMTP id p73LjjYO014360 for ; Wed, 3 Aug 2011 14:45:45 -0700 Received: from gwb15 (gwb15.prod.google.com [10.200.2.15]) by hpaq6.eem.corp.google.com with ESMTP id p73LjWfh024883 (version=TLSv1/SSLv3 cipher=RC4-SHA bits=128 verify=NOT) for ; Wed, 3 Aug 2011 14:45:44 -0700 Received: by gwb15 with SMTP id 15so738611gwb.41 for ; Wed, 03 Aug 2011 14:45:44 -0700 (PDT) Received: by 10.151.51.16 with SMTP id d16mr1307943ybk.16.1312407943969; Wed, 03 Aug 2011 14:45:43 -0700 (PDT) MIME-Version: 1.0 Received: by 10.151.51.16 with SMTP id d16mr1307933ybk.16.1312407943773; Wed, 03 Aug 2011 14:45:43 -0700 (PDT) Received: by 10.151.101.7 with HTTP; Wed, 3 Aug 2011 14:45:43 -0700 (PDT) In-Reply-To: <20110803210614.GE22893@atrey.karlin.mff.cuni.cz> References: <4E25F810.6050904@redhat.com> <4E2D4425.9040104@gnu.org> <20110803210614.GE22893@atrey.karlin.mff.cuni.cz> Date: Wed, 03 Aug 2011 21:46:00 -0000 Message-ID: Subject: Re: [RFC] Remove -freorder-blocks-and-partition From: Xinliang David Li To: Jan Hubicka Cc: Paolo Bonzini , Richard Henderson , "gcc@gcc.gnu.org" Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable X-System-Of-Record: true X-IsSubscribed: yes Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org X-SW-Source: 2011-08/txt/msg00076.txt.bz2 On Wed, Aug 3, 2011 at 2:06 PM, Jan Hubicka wrote: >> In xalancbmk, with the partition option, most of object files have >> nonzero size cold sections generated. The text size of the binary is >> increased to 3572728 bytes from 3466790 bytes. =A0Profiling the program >> using the training input shows the following differences. With >> partitioning, number of executed branch instructions slightly >> increases, but itlb misses and icache load misses are significantly >> lower compared with the binary without partitioning. >> >> >> David >> >> With partition: >> ----------------- >> =A0 =A053654937239 =A0branches >> =A0 =A0 =A0 306751458 =A0L1-icache-load-misses >> =A0 =A0 =A0 =A0 8146112 =A0iTLB-load-misses > > Note that I was also planning for some time to introduce notion of provab= ly cold > stuff into our branch prediction heurstics. I.e. code leading to aborts, = eh etc no-return attribute is looked at by static profile estimation pass. Is the attribute (definitely not returning) properly propagated to the callers (wrappers of exit, etc)? David > that can be then offlined even w/o profile feedback and could perhaps help > to large apps. > (also the whole pass should be more effective with larger testcases, SPEC= 2k6 is slowly > becoming a small one) > > Honza >