From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 14434 invoked by alias); 9 Nov 2007 22:46:19 -0000 Received: (qmail 14418 invoked by uid 22791); 9 Nov 2007 22:46:18 -0000 X-Spam-Check-By: sourceware.org Received: from rn-out-0910.google.com (HELO rn-out-0102.google.com) (64.233.170.184) by sourceware.org (qpsmtpd/0.31) with ESMTP; Fri, 09 Nov 2007 22:46:16 +0000 Received: by rn-out-0102.google.com with SMTP id s28so501552rnb for ; Fri, 09 Nov 2007 14:46:14 -0800 (PST) Received: by 10.142.232.20 with SMTP id e20mr332309wfh.1194648373724; Fri, 09 Nov 2007 14:46:13 -0800 (PST) Received: by 10.142.84.21 with HTTP; Fri, 9 Nov 2007 14:46:13 -0800 (PST) Message-ID: <571f6b510711091446r16bff2fbu2f720d16367452e6@mail.gmail.com> Date: Fri, 09 Nov 2007 23:59:00 -0000 From: "Steven Bosscher" To: gcc-patches Subject: [PATCH] gcse.c: don't compute a mapping from CUID to INSN MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline 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 X-SW-Source: 2007-11/txt/msg00551.txt.bz2 Hi, I've long lost the illusion that anyone here cares one serious bit about compile time. But having loops over all insns just for the sake of it is surely too much even for the most stubborn pass-adding, compile time increasing gcc hacker around here? Thus. Don't do useless work in gcse.c. This saves 2 passes over all insns with -fgcse, and even 3 with -Os. I doubt this will have any measurable impact, but one has to start somewhere. Patch is untested beyond "it compiles", but that's really sufficient in this case anyway :-) Gr. Steven * gcse.c (CUID_INSN): Remove. (cuid_insn): Ditto. (alloc_gcse_mem): Don't allocate cuid_insn. (free_gcse_mem): Don't free cuid_insn.