From owner-freebsd-hackers Mon Aug 11 17:51:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA12885 for hackers-outgoing; Mon, 11 Aug 1997 17:51:09 -0700 (PDT) Received: from ducky.net (gate.ducky.net [198.145.101.253]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id RAA12860; Mon, 11 Aug 1997 17:51:03 -0700 (PDT) Received: from localhost.ducky.net (localhost.ducky.net [127.0.0.1]) by ducky.net (8.6.12/8.6.12) with SMTP id RAA04916; Mon, 11 Aug 1997 17:50:45 -0700 Message-Id: <199708120050.RAA04916@ducky.net> X-Authentication-Warning: ducky.net: Host localhost.ducky.net didn't use HELO protocol To: Curt Sampson cc: freebsd-hackers@FreeBSD.ORG, freebsd-hardware@FreeBSD.ORG Subject: Re: question about "ed" driver performance on ASUS SP3G & 486DX4/100 In-reply-to: Your message of "Mon, 11 Aug 1997 08:48:02 PDT." Date: Mon, 11 Aug 1997 17:50:45 -0700 From: Mike Haertel Sender: owner-freebsd-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk >> I tried 16-bit copies; they were a little slower (about 16 and 18 >> seconds on the P5-90 and 486-100 respectively). > >I'd be interested to see what you get with 8-bit copies. Did you >drop the board into 16-bit mode before doing your 16-bit reads, >BTW? I'm not sure if it makes a difference on reads, but it certainly >does on writes. 8 bit copies were 1.5 times as slow as 16-bit copies--about .65 megabytes/sec. I tried both rep movsb and manually unrolled movb's. At this point, I've concluded the board is probably at fault...