From owner-freebsd-current Wed Dec 17 10:13:57 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id KAA29049 for current-outgoing; Wed, 17 Dec 1997 10:13:57 -0800 (PST) (envelope-from owner-freebsd-current) Received: from ns1.yes.no (ns1.yes.no [195.119.24.10]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id KAA29030; Wed, 17 Dec 1997 10:13:38 -0800 (PST) (envelope-from eivind@bitbox.follo.net) Received: from bitbox.follo.net (bitbox.follo.net [194.198.43.36]) by ns1.yes.no (8.8.7/8.8.7) with ESMTP id SAA23394; Wed, 17 Dec 1997 18:13:29 GMT Received: (from eivind@localhost) by bitbox.follo.net (8.8.6/8.8.6) id TAA19921; Wed, 17 Dec 1997 19:13:25 +0100 (MET) To: dyson@FreeBSD.ORG Cc: garbanzo@hooked.net (Alex), current@FreeBSD.ORG Subject: Re: Pentium optimizations References: <199712170623.BAA00476@dyson.iquest.net> From: Eivind Eklund Date: 17 Dec 1997 19:13:24 +0100 In-Reply-To: "John S. Dyson"'s message of Wed, 17 Dec 1997 01:23:22 -0500 (EST) Message-ID: <86pvmvaldn.fsf@bitbox.follo.net> Lines: 19 X-Mailer: Gnus v5.4.52/XEmacs 20.2 Sender: owner-freebsd-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk "John S. Dyson" writes: > I do have some PPro mods, and they appear to help performance on > average. The PPro is a really wierd creature (like the K6.) The > darned processor does so much optimization, it appears to be > insensitive to code mods. There are areas of reasonable payoffs, > and lots of "obvious" optimizations that end up being neutral. I was working with optimizing assembly code for the PPro a year ago. My experience was that modifying code seldom mattered, except for alignement. Making the tight loops hit 16-byte boundaries roughly doubled the speed. No other approach made a significant difference, AFAIR (I just supplied ideas and had another programmer implement them). All pairing happened automatically, and touching the cache to make it pre-fetch etc didn't help at all. Eivind.