From owner-freebsd-current@FreeBSD.ORG Wed Aug 11 21:23:46 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 86CB616A58B; Wed, 11 Aug 2004 21:23:46 +0000 (GMT) Received: from gw.catspoiler.org (217-ip-163.nccn.net [209.79.217.163]) by mx1.FreeBSD.org (Postfix) with ESMTP id 9815F43D1F; Wed, 11 Aug 2004 21:23:45 +0000 (GMT) (envelope-from truckman@FreeBSD.org) Received: from FreeBSD.org (mousie.catspoiler.org [192.168.101.2]) by gw.catspoiler.org (8.12.11/8.12.11) with ESMTP id i7BLNZfX018166; Wed, 11 Aug 2004 14:23:39 -0700 (PDT) (envelope-from truckman@FreeBSD.org) Message-Id: <200408112123.i7BLNZfX018166@gw.catspoiler.org> Date: Wed, 11 Aug 2004 14:23:35 -0700 (PDT) From: Don Lewis To: rwatson@FreeBSD.org In-Reply-To: MIME-Version: 1.0 Content-Type: TEXT/plain; charset=us-ascii cc: freebsd-current@FreeBSD.org cc: mb@imp.ch Subject: Re: SCHEDULE and high load situations X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 11 Aug 2004 21:23:46 -0000 On 11 Aug, Robert Watson wrote: > > On Wed, 11 Aug 2004, Martin Blapp wrote: > >> During load tests with both SCHED4BSD and SCHEDULE I found that the >> latest SCHEDULE version cannot have a load over 2-3. I stress tested the >> server really a lot, but about 500 processes did sleep and did not got >> scheduled and only 2 of 500 were run and got about 50% of CPU time. >> >> Is this a known problem ? > > I've found that for throughput oriented workloads, 4BSD substantially > outperforms ULE, but I haven't tried it with Jeff's latest set of patches > (committed a day or two ago). You don't mention if your box is SMP, btw > -- I've noticed some load balancing problems with ULE previously, but > haven't checked if they were resolved. Anecdotal opinion seems generally > to be that interactivity is observably better with ULE than 4BSD, but that > 4BSD appears to do a better job under load. I've noticed in fork()/exec() intensive loads like buildworld and building ports on a UP box that ULE favors a CPU-bound but niced process like setiathome over the software build. Even the longer c++ compile steps seem to get less CPU time, than the niced CPU-bound process, at least according to top. Buildworld times with ULE on my Athlon XP box are about 145 minutes with ULE and 82 minutes with 4BSD when competing with setiathome. Top shows setiathome consistently getting about 55% of the CPU with ULE. Without setiathome running, buildworld takes about 65 minutes.