From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 31770 invoked by alias); 5 Jan 2003 14:24:46 -0000 Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org Received: (qmail 31763 invoked from network); 5 Jan 2003 14:24:45 -0000 Received: from unknown (HELO nile.gnat.com) (205.232.38.5) by 209.249.29.67 with SMTP; 5 Jan 2003 14:24:45 -0000 Received: by nile.gnat.com (Postfix, from userid 338) id 3F706F2E1C; Sun, 5 Jan 2003 09:24:33 -0500 (EST) To: dewar@gnat.com, lord@emf.net Subject: Re: An unusual Performance approach using Synthetic registers Cc: gcc@gcc.gnu.org, ja_walker@earthlink.net Message-Id: <20030105142433.3F706F2E1C@nile.gnat.com> Date: Sun, 05 Jan 2003 15:47:00 -0000 From: dewar@gnat.com (Robert Dewar) X-SW-Source: 2003-01/txt/msg00217.txt.bz2 > Otherwise, what you lose for locals/args (the dominant case) will > probably exceed what you gain for other values. Actually I think you will break even most of the time and generate essentially identical code. You will end up saying, "great, I can keep this local variable Q in a register, I don't need to store it in the stack frame, but then the register turns out to be a SR, and in fact it is right back there in the stack frame, with identical instructions used to access it.