From owner-cvs-all@FreeBSD.ORG Tue Oct 30 17:54:49 2007 Return-Path: Delivered-To: cvs-all@FreeBSD.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id A9A4516A41B; Tue, 30 Oct 2007 17:54:49 +0000 (UTC) (envelope-from kris@FreeBSD.org) Received: from weak.local (pointyhat.freebsd.org [IPv6:2001:4f8:fff6::2b]) by mx1.freebsd.org (Postfix) with ESMTP id 8C7FF13C4B5; Tue, 30 Oct 2007 17:54:48 +0000 (UTC) (envelope-from kris@FreeBSD.org) Message-ID: <47276FEB.2030202@FreeBSD.org> Date: Tue, 30 Oct 2007 18:54:51 +0100 From: Kris Kennaway User-Agent: Thunderbird 2.0.0.6 (Macintosh/20070728) MIME-Version: 1.0 To: David Xu References: <200710292101.l9TL1mAE049561@repoman.freebsd.org> <47268F17.1000106@freebsd.org> <47269AD0.3080906@freebsd.org> <4726EEFB.8030309@FreeBSD.org> <4726F3C7.2060506@freebsd.org> <472701C9.4020208@FreeBSD.org> <47272E64.4080607@126.com> In-Reply-To: <47272E64.4080607@126.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Daniel Eischen , David Xu , cvs-all@FreeBSD.org Subject: Re: cvs commit: src/lib/libthr/thread thr_mutex.c src/lib/libkse/thread thr_mutex.c src/include pthread.h X-BeenThere: cvs-all@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: CVS commit messages for the entire tree List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 30 Oct 2007 17:54:49 -0000 David Xu wrote: > Kris Kennaway wrote: > > >>> My last commit improves mysql select-smack benchmark on 4-core xeon from >>> 48000 queries/s to 70000 queries/s, so my work is alternative way >> >> No, that is an orthogonal issue that (after measurement) does not >> solve the same problem that is addressed by this change. I'd be happy >> to discuss it with you in more detail if you are interested. We could >> also discuss the fact that super-smack is a questionable target to be >> optimizing because the main performance problems seem to be from a >> very poor benchmark design. >> >> I claim that real-world applications do not commonly do I/O in units >> of 1 byte :) > > I'd rather to believe it is scheduling problem, spin loops does not help > it, but sched_yield() loops does help it a lot. Sure, maybe. It's still a completely orthogonal problem. Kris