From owner-freebsd-questions Tue Dec 10 10:35:25 1996 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.4/8.8.4) id KAA20195 for questions-outgoing; Tue, 10 Dec 1996 10:35:25 -0800 (PST) Received: from Campino.Informatik.RWTH-Aachen.DE (campino.Informatik.RWTH-Aachen.DE [137.226.116.240]) by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id KAA20137 for ; Tue, 10 Dec 1996 10:33:55 -0800 (PST) Received: from gilberto.physik.rwth-aachen.de (gilberto.physik.rwth-aachen.de [137.226.31.2]) by Campino.Informatik.RWTH-Aachen.DE (RBI-Z-5/8.6.12) with ESMTP id TAA15011; Tue, 10 Dec 1996 19:35:17 +0100 (MET) Received: (from kuku@localhost) by gilberto.physik.rwth-aachen.de (8.8.3/8.6.9) id TAA16437; Tue, 10 Dec 1996 19:47:40 +0100 (MET) From: Christoph Kukulies Message-Id: <199612101847.TAA16437@gilberto.physik.rwth-aachen.de> Subject: Re: FreeBSD on a PPRO In-Reply-To: <3.0.32.19961210182048.00ba6680@istudio.no> from Simon Lindgren at "Dec 10, 96 06:20:49 pm" To: lindgren@istudio.no (Simon Lindgren) Date: Tue, 10 Dec 1996 19:47:40 +0100 (MET) Cc: freebsd-questions@FreeBSD.ORG Reply-To: Christoph Kukulies X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-questions@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Does FreeBSD (2.1.6R) take full potential of a Pentium PRO? > Will there be a significant speed-increase when upgrading from a > regular pentium of same speeds? > > Some insight would be appreciated. We have P5/166 here and one P6/200 and the first results when people compiled their large physics programs (which are doing very little disk i/o) were disappointing. They ran some Monte Carlos over 20 hours and it turned out that one P5/166 was slightly faster than tha P6/200 (!). I then brought pgcc on the PPRO and recompiled all libm (msun) with -O2 -ffast-math -m486 and linked the program statically. With this program then we went on the two machines and the PPro/200 (256K cache) turned out to be 1.5 time faster than th P5/166. I wonder where I might squeeze out the last microseconds. Perhaps carefully selecting the bzero optimization? > > > > ------------------------------------------------------ > Simon Lindgren | Webarkitekt/Produktsjef > lindgren@istudio.no | Internett Studio AS > http://www.sn.no/~lindgren | http://www.istudio.no/ > ------------------------------------------------------ > > --Chris Christoph P. U. Kukulies kuku@gil.physik.rwth-aachen.de