From owner-freebsd-hackers Wed Jun 7 12:49:52 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id MAA02899 for hackers-outgoing; Wed, 7 Jun 1995 12:49:52 -0700 Received: from uivlsi.csl.uiuc.edu (uivlsi.csl.uiuc.edu [128.174.57.133]) by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id MAA02893 for ; Wed, 7 Jun 1995 12:49:50 -0700 Received: by uivlsi.csl.uiuc.edu id AA25272 (5.67b/IDA-1.3.4 for freebsd-hackers@freebsd.org); Wed, 7 Jun 1995 14:49:49 -0500 Date: Wed, 7 Jun 1995 14:49:49 -0500 From: Terry Lee Message-Id: <199506071949.AA25272@uivlsi.csl.uiuc.edu> To: freebsd-hackers@freebsd.org Subject: WD driver serendipity Sender: hackers-owner@freebsd.org Precedence: bulk I mailed the following to freebsd-hackers a couple of days ago, but it never showed up. Anyway, amid the usual "this doesn't work" reports during alpha testing (or any other time it seems :=) ), I'd like to report something working that I did not expect. Isn't ST506 multi-block transfer capability available on only IDE disks? I have an UltraStor 12F ESDI disk controller. After reading that 32-bit and multi-block transfers are now supported by the wd driver, I decided to enable "flags wdc0 0x80ff80ff" thinking, "bwa ha ha, this could be fun!" To my surprise, the wd driver reported "multi-block-8" and the CPU usage of the wd driver decreased by at least 10% on my 486DX-40 system. Thanks for the great work! If 32-bit and multi-block transfers don't cause problems, I hope they will be enabled by default in 2.1 since I suspect not everyone is aware of them. Has multi-block transfer been available on ST506 disk controllers all along? If so, has it taken that long (after IDE disks were introduced) for the BIOS manufacturers to support this capability? Terry Lee terry@uivlsi.csl.uiuc.edu