From mboxrd@z Thu Jan 1 00:00:00 1970 From: "J. R. Okajima" Subject: Re: Big git diff speedup by avoiding x86 "fast string" memcmp Date: Mon, 13 Dec 2010 16:29:11 +0900 Message-ID: <9580.1292225351@jrobl> References: <20101209070938.GA3949@amd> <19324.1291990997@jrobl> <20101213014553.GA6522@amd> Return-path: Received: from mtoichi14.ns.itscom.net ([219.110.2.184]:39002 "EHLO mtoichi14.ns.itscom.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752765Ab0LMHh2 (ORCPT ); Mon, 13 Dec 2010 02:37:28 -0500 In-Reply-To: <20101213014553.GA6522@amd> Sender: linux-arch-owner@vger.kernel.org List-ID: To: Nick Piggin Cc: Linus Torvalds , linux-arch@vger.kernel.org, x86@kernel.org, linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org Nick Piggin: > It's not scaling but just single threaded performance. gcc turns memcmp > into rep cmp, which has quite a long latency, so it's not appripriate > for short strings. Honestly speaking I doubt how this 'long *' approach is effective (Of course it never means that your result (by 'char *') is doubtful). But is the "rep cmp has quite a long latency" issue generic for all x86 architecture, or Westmere system specific? J. R. Okajima