From owner-freebsd-hackers Tue Aug 13 16:38:15 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA20332 for hackers-outgoing; Tue, 13 Aug 1996 16:38:15 -0700 (PDT) Received: from mail.think.com (Mail1.Think.COM [131.239.33.245]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id QAA20326 for ; Tue, 13 Aug 1996 16:38:12 -0700 (PDT) Received: from Early-Bird.Think.COM (Early-Bird-1.Think.COM [131.239.146.105]) by mail.think.com (8.7.5/m3) with ESMTP id TAA03344; Tue, 13 Aug 1996 19:38:03 -0400 (EDT) Received: from compound.Think.COM ([206.147.16.34]) by Early-Bird.Think.COM (8.7.5/e1) with ESMTP id TAA04117; Tue, 13 Aug 1996 19:37:59 -0400 (EDT) Received: (from alk@localhost) by compound.Think.COM (8.7.5/8.7.3) id SAA09454; Tue, 13 Aug 1996 18:39:09 -0500 (CDT) Date: Tue, 13 Aug 1996 18:39:09 -0500 (CDT) From: Tony Kimball Message-Id: <199608132339.SAA09454@compound.Think.COM> To: michaelv@HeadCandy.com Cc: dennis@etinc.com, hackers@freebsd.org Subject: Re: FreeBSD vs. NT Stability Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Quoth Michael L. VanLoon on Mon, 12 August: : >This is all very nice, but on a macro level NT has two very obvious : >problems to consider: : >1) Its rather new... : >2) It was written by Microsoft : : And, how do these affect NT's performance? (Or, threading vs. process : context switching, in general?) : Well, it means that you can't run the same benchmarks for one thing: It won't run standard thread code, nor will it run standard multiprocess code. In fact NT's performance metric for most classic benchmarks will be precisely 0. It also means that one is well-justified in inferring that system uptime will be short enough so that many performance measurements (those relating to very long running jobs) will be mooted by stability problems. Actually, my impression is that 4.0 is quite an improvement over past incarnations in terms of both robustness and efficiency, although it does suck memory badly, so the historically justified induction may break down for NT 4.0. Such benchmarks seem of questionable relevance anyhow -- NT is not selected for performance -- or for compatibility in environments where NetWare is absent. NT is selected to run a GUI, and to run specific commercial applications which aren't generally available on unix or bsd systems anyhow, so unix and bsd get a big goose-egg performance metric.