From owner-freebsd-stable@FreeBSD.ORG Sat Dec 17 17:33:27 2011 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id E75211065670 for ; Sat, 17 Dec 2011 17:33:27 +0000 (UTC) (envelope-from george+freebsd@m5p.com) Received: from mailhost.m5p.com (ip-3-2-0-2.r20.asbnva02.us.ce.gin.ntt.net [IPv6:2001:418:0:5000::16]) by mx1.freebsd.org (Postfix) with ESMTP id 7F2228FC0A for ; Sat, 17 Dec 2011 17:33:27 +0000 (UTC) Received: from wonderland.m5p.com (wonderland.m5p.com [IPv6:2001:418:3fd::19]) by mailhost.m5p.com (8.14.4/8.14.4) with ESMTP id pBHHXLHw054328; Sat, 17 Dec 2011 12:33:26 -0500 (EST) (envelope-from george+freebsd@m5p.com) Message-ID: <4EECD261.2080208@m5p.com> Date: Sat, 17 Dec 2011 12:33:21 -0500 From: George Mitchell User-Agent: Mozilla/5.0 (X11; FreeBSD amd64; rv:8.0) Gecko/20111127 Thunderbird/8.0 MIME-Version: 1.0 To: Oliver Pinter References: <4EE1EAFE.3070408@m5p.com> <4EE2AE64.9060802@m5p.com> <4EE88343.2050302@m5p.com> <4EE933C6.4020209@zedat.fu-berlin.de> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Greylist: Sender passed SPF test, not delayed by milter-greylist-4.2.7 (mailhost.m5p.com [IPv6:2001:418:3fd::f7]); Sat, 17 Dec 2011 12:33:26 -0500 (EST) X-Scanned-By: MIMEDefang 2.72 on IPv6:2001:418:3fd::f7 Cc: freebsd-stable@freebsd.org Subject: Re: SCHED_ULE should not be the default X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 17 Dec 2011 17:33:28 -0000 On 12/14/11 21:05, Oliver Pinter wrote: > [...] > Hi! > > Can you try with this settings: > > op@opn ~> sysctl kern.sched. > kern.sched.cpusetsize: 8 > kern.sched.preemption: 0 > kern.sched.name: ULE > kern.sched.slice: 13 > kern.sched.interact: 30 > kern.sched.preempt_thresh: 224 > kern.sched.static_boost: 152 > kern.sched.idlespins: 10000 > kern.sched.idlespinthresh: 16 > kern.sched.affinity: 1 > kern.sched.balance: 1 > kern.sched.balance_interval: 133 > kern.sched.steal_htt: 1 > kern.sched.steal_idle: 1 > kern.sched.steal_thresh: 1 > kern.sched.topology_spec: > > 0, 1 > > > 0, 1 > > > > > [...] Sorry I didn't try this earlier, but I had time this morning. Apparently you can't change kern.sched.preemption without recompiling, so I did that. It didn't help, and subjectively it made interactive performance worse. I changed preempt_thresh and observed no difference. There were only a couple of small differences between your other settings and the 9.0-PRERELEASE defaults. Summing up for the record, in my original test: 1. It doesn't matter whether X is running or not. 2. The problem is not limited to two or fewer CPUs. (It also happens for me on a six-CPU system.) 3. It doesn't require nCPU + 1 compute-bound processes, just nCPU. With nCPU compute-bound processes running, with SCHED_ULE, any other process that is interactive (which to me means frequently waiting for I/O) gets ABYSMAL performance -- over an order of magnitude worse than it gets with SCHED_4BSD under the same conditions. -- George