Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!usc!rutgers!dptg!att!drutx!druwy!dlm From: dlm@druwy.ATT.COM (Dan Moore) Newsgroups: comp.sys.atari.st Subject: Re: ATTENTION SPECTRE 128 OWNERS with fast SCSI drives Message-ID: <4364@druwy.ATT.COM> Date: 29 Sep 89 15:00:54 GMT References: <890926.11092003.048625@SFA.CP6> Organization: AT&T, Denver, CO Lines: 35 in article <890926.11092003.048625@SFA.CP6>, Z4648252@SFAUSTIN.BITNET (Z4648252) says: > As David Small said, some fast SCSI drives will require the slow > SCSI option to be selected. Otherwise, one is liable to get read/write > errors, etc. With the slow SCSI option, the problems are cleared and > [ ... ] > I'm using the SeaGate 296n. Yeah, I've been reading all the talk > about interweave problems and ROM bugs, but even at 2:1 interweave, ^^^^^^^^^^ should be interleave > the mechanism is still fast. It is so fast, that slow SCSI option has > to be selected. For my system, that slows things down so much, across > the board, that my hard drive I/O results in access slower than my > creaking SH 204 hard drive!!! You have the problem backwards. Some SCSI drives (mostly Seagate SCSI drives) are too slow, not too fast. These drives do not correctly respond to a new SCSI command if it is received "too soon" after the completion of a previous command. (eg. issue a read command for 10 sectors and as soon as that read completes issue another for 8 sectors, the second read command will fail on drives that have this bug.) There are two different "speed" issues involved. One is the data transfer rate which is set by the drive interleave. The second is the speed at which the drive can accept SCSI commands. A "fast" drive that transfers data quickly may or may not have problems with accepting SCSI commands quickly. Dan Moore AT&T Bell Labs Denver dlm@druwy.ATT.COM