From owner-freebsd-current Thu Oct 31 21:21:18 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id VAA17137 for current-outgoing; Thu, 31 Oct 1996 21:21:18 -0800 (PST) Received: from mx.serv.net (mx.serv.net [199.201.191.10]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id VAA17132 for ; Thu, 31 Oct 1996 21:21:15 -0800 (PST) Received: from MindBender.serv.net by mx.serv.net (8.7.5/SERV Revision: 2.30) id VAA24592; Thu, 31 Oct 1996 21:21:17 -0800 (PST) Received: from localhost.HeadCandy.com (michaelv@localhost.HeadCandy.com [127.0.0.1]) by MindBender.serv.net (8.7.5/8.7.3) with SMTP id VAA19832; Thu, 31 Oct 1996 21:21:02 -0800 (PST) Message-Id: <199611010521.VAA19832@MindBender.serv.net> X-Authentication-Warning: MindBender.serv.net: Host michaelv@localhost.HeadCandy.com [127.0.0.1] didn't use HELO protocol To: jlemon@americantv.com (Jonathan Lemon) cc: roberto@keltia.freenix.fr (Ollivier Robert), current@freebsd.org Subject: Re: 2.1.5r -> current upgrade In-reply-to: Your message of Thu, 31 Oct 96 14:39:04 -0600. <199610312039.UAA03047@right.PCS> Date: Thu, 31 Oct 1996 21:21:02 -0800 From: "Michael L. VanLoon -- HeadCandy.com" Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >Ollivier Robert writes: >> According to Warner Losh: >> > On my 486DX2/66 + VLB Ultrastor controller + 32M it is between 9 and >> > 10 hours. I'd really like to see an 82 minute make world, but I don't >> I was seeing about 9h50 on my 486DX33 + 32 MB + Buslogic 747S EISA. I'm >> now at less than 4 hours on my (now upgraded) 486DX4/100 + 32 MB + a faster >> drive and /usr/src & /usr/obj on different SCSI controllers (the other one >> is an AHA-1740A). jlemon@americantv.com (Jonathan Lemon) writes: >Does this mean that the biggest performance boost comes from adding an >additional controller? Hm. I was thinking my poor little DX2/66 was >mainly CPU bound. I'll have to try putting in a spare 1542 into my EISA >machine and seeing if that makes a difference. In my experience, you will never do any better with a 486. It *is* CPU bound. Or at least a 486 will have much higher latency (turn-around from servicing a completed SCSI task to queuing another task). Less than four hours sounds a little *too* good for a *full* make world on any 486. Are you sure you didn't have some things turned off (like building profiled libs, or formatting man pages (I'm not even sure FreeBSD does that, like NetBSD does))? The world trees have definitely diverged, but I don't think the times will be that far off. My relative times for my personal "make world" script (I call it "doall") on NetBSD, which does a few things redundantly, then clean builds a kernel, are as follows -- this is a full clean build, remaking all dependencies, reformatting all man pages, building all versions of the libraries, etc.: 486-133MHz: A little over 6 hours Pentium-120MHz: 3:15 Pentium Pro-200MHz: 1:21 486: NICE SuperEISA 486 motherboard, AMD 5x86 133MHz, 512K L2 write-back cache, 24MB RAM, BusLogic BT747s Fast/Narrow EISA SCSI controller, 2 - HP 1GB Fast/Narrow SCSI-2 hard drives striped as a ccd P5: Asus P/I-P55TP4N (Triton-1) motherboard, Intel Pentium 120MHz, 512K L2 pipeline-burst cache, 64MB EDO RAM, Adaptec 2940UW Ultra/Wide PCI SCSI controller, same hard drives P6: Asus P/I-P6NP5 (Natoma) motherboard, Intel Pentium Pro 200MHz 256K L2 cache, 64MB EDO RAM, same SCSI controller and drives ----------------------------------------------------------------------------- Michael L. VanLoon michaelv@MindBender.serv.net --< Free your mind and your machine -- NetBSD free un*x >-- NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... NetBSD ports in progress: PICA, others... -----------------------------------------------------------------------------