From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 18658 invoked by alias); 19 Dec 2007 13:50:37 -0000 Received: (qmail 18624 invoked by uid 48); 19 Dec 2007 13:50:26 -0000 Date: Wed, 19 Dec 2007 13:50:00 -0000 Message-ID: <20071219135026.18623.qmail@sourceware.org> X-Bugzilla-Reason: CC References: Subject: [Bug rtl-optimization/17236] inefficient code for long long multiply on x86 In-Reply-To: Reply-To: gcc-bugzilla@gcc.gnu.org To: gcc-bugs@gcc.gnu.org From: "bonzini at gnu dot org" Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org X-SW-Source: 2007-12/txt/msg01814.txt.bz2 ------- Comment #25 from bonzini at gnu dot org 2007-12-19 13:50 ------- Note that "fwprop" is not an exact term, because there *is* a memory load in each multiplication, and propagating a second memory operand will create an invalid insn. You may try to add a split from reg=op(mem1, mem2) to reg=mem1;reg=reg*mem2, but anyway, it won't help propagating into the mull, since it is not LOG_LINKS-related to the load. The problem is that we end up with having to reload *anyway*. If we had a way to express the relationship between operands 0 and 1 of the multiplication *before reload*, probably the RTL optimizations could do much more. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17236