From owner-freebsd-current Fri Jun 28 12:58:43 2002 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 6F8ED37B40F for ; Fri, 28 Jun 2002 12:58:03 -0700 (PDT) Received: from magic.adaptec.com (magic.adaptec.com [208.236.45.80]) by mx1.FreeBSD.org (Postfix) with ESMTP id 4865043E1B for ; Fri, 28 Jun 2002 12:57:48 -0700 (PDT) (envelope-from Scott_Long@adaptec.com) Received: from redfish.adaptec.com (redfish.adaptec.com [162.62.50.11]) by magic.adaptec.com (8.10.2+Sun/8.10.2) with ESMTP id g5SJv3E20779; Fri, 28 Jun 2002 12:57:03 -0700 (PDT) Received: from btc.btc.adaptec.com (btc.btc.adaptec.com [10.100.0.52]) by redfish.adaptec.com (8.8.8+Sun/8.8.8) with ESMTP id MAA21762; Fri, 28 Jun 2002 12:56:58 -0700 (PDT) Received: from btcexc01.btc.adaptec.com (btcexc01 [10.100.0.23]) by btc.btc.adaptec.com (8.8.8+Sun/8.8.8) with ESMTP id NAA15005; Fri, 28 Jun 2002 13:56:55 -0600 (MDT) Received: by btcexc01.btc.adaptec.com with Internet Mail Service (5.5.2653.19) id ; Fri, 28 Jun 2002 13:56:56 -0600 Message-ID: <2C7CBDC6EA58D6119E4A00065B3A24CB0463ED@btcexc01.btc.adaptec.com> From: "Long, Scott" To: "'Terry Lambert'" , Matthew Dillon Cc: Julian Elischer , FreeBSD current users , "'kirk@mckusick.com'" Subject: RE: something funny with soft updates? Date: Fri, 28 Jun 2002 13:56:54 -0600 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Terry Lambert wrote: > Matthew Dillon wrote: > > :there may not be anything wrong but I expected more improvement. > > :I remember that when I first committed softupdates it made > a noticable > > :difference.... > > > > Julian, please note that a buildworld does *NOT* > seriously exercise > > the filesystem. I/O ops are mostly reads. The only area where > > softupdates might have a major effect would be the rm -rf of the > > object tree at the beginning. > > I still want to know why his KSE numbers were better than the > non-KSE numbers when KSE was not using soft updates, and they > are the same, now that he *is* using soft updates. > > In general, there is (was?) a positive difference in buildworld > times when using soft updates, and Julian seems to have now > demonstrated that this is no longer the case. > > It would be interesting to find out what broke and who broke it, > which is what I think he's asking, and recover the speedup, and > find out if it's all in the object tree delete, or not (I think > that Julian starts without an object tree for this masurement, > right?). > > Maybe the difference in speed is related to the UFS2 commit? > > -- Terry > I'm seeing similar weirdness in 4.6-STABLE. It seems to be especially pronounced on RAID arrays (irregardless of hardware vs software RAID). Memory size doesn't seem to make a difference, as an unloaded 2GB machine of mine behaves in exactly the same way. Scott To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message