From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 16851 invoked by alias); 23 Feb 2008 12:49:19 -0000 Received: (qmail 16839 invoked by uid 22791); 23 Feb 2008 12:49:18 -0000 X-Spam-Check-By: sourceware.org Received: from wf-out-1314.google.com (HELO wf-out-1314.google.com) (209.85.200.170) by sourceware.org (qpsmtpd/0.31) with ESMTP; Sat, 23 Feb 2008 12:49:01 +0000 Received: by wf-out-1314.google.com with SMTP id 28so451770wfc.14 for ; Sat, 23 Feb 2008 04:48:59 -0800 (PST) Received: by 10.142.187.2 with SMTP id k2mr230413wff.25.1203770939433; Sat, 23 Feb 2008 04:48:59 -0800 (PST) Received: by 10.142.100.19 with HTTP; Sat, 23 Feb 2008 04:48:59 -0800 (PST) Message-ID: <84fc9c000802230448u5086e0b2w311e5bace92b6861@mail.gmail.com> Date: Sat, 23 Feb 2008 12:51:00 -0000 From: "Richard Guenther" To: "Diego Novillo" Subject: Re: [PATCH][4.4] Add a global DECL_UID -> tree mapping Cc: "Andrew MacLeod" , "Richard Guenther" , gcc-patches@gcc.gnu.org In-Reply-To: <47BF5EEB.2030107@google.com> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline References: <47BF1BCF.8010300@redhat.com> <47BF5EEB.2030107@google.com> X-IsSubscribed: yes 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: 2008-02/txt/msg01016.txt.bz2 On Sat, Feb 23, 2008 at 12:46 AM, Diego Novillo wrote: > On 2/22/08 2:00 PM, Andrew MacLeod wrote: > > > This oughta fix the odd compile time hit we have when the hash table > > doesn't perform well too eh? I remember looking at something a few weeks > > ago where most of the operand scan time (which was needlessly excessive) > > was almost all in referenced_var hash table lookups. getting rid of > > that hash lookup is a good thing. Though we replace it with a hash lookup on the global table. > Indeed. I like this change quite a bit. Thanks for working on this > Richard. It might also prove useful for IPA analysis where you now can track vars with just their UIDs (rather than function + UID). Richard.