Return to BSD News archive
Xref: sserve comp.os.linux:39999 comp.os.386bsd.questions:2627 comp.unix.pc-clone.32bit:2946 Newsgroups: comp.os.linux,comp.os.386bsd.questions,comp.unix.pc-clone.32bit Path: sserve!newshost.anu.edu.au!munnari.oz.au!spool.mu.edu!uwm.edu!ux1.cso.uiuc.edu!moe.ksu.ksu.edu!engr.uark.edu!rzrbyte.fay.ar.us!tep From: tep@rzrbyte.fay.ar.us (Tim Peoples) Subject: Re: IDE vs SCSI and performance References: <C72CAw.B47@sugar.NeoSoft.COM> <1993May17.190905.3462@gandalf.ca> <C78rG8.707@rex.uokhsc.edu> <1tftbuINNn9l@rave.larc.nasa.gov> Organization: RazorByte Software Systems, Ltd. Date: Thu, 20 May 1993 23:44:28 GMT Message-ID: <1993May20.234428.12715@rzrbyte.fay.ar.us> Lines: 55 jcburt@gatsibm.larc.nasa.gov () writes: >In article <C78rG8.707@rex.uokhsc.edu> benjamin-goldsteen@uokhsc.edu writes: >>ykhan@gandalf.ca (Yousuf Khan) writes: >> >>>In <C72CAw.B47@sugar.NeoSoft.COM> peter@NeoSoft.com (Peter da Silva) writes: >> >>>>I was talking to a fellow in a computer store the other day, and he was >>>>insisting that he was getting 2.5 MB/s on his IDE drives under AmigaOS, >>>>over twice what he got with SCSI. >> > >[...stuff deleted...] > >>>>I found that hard to beleive... I suspected that his benchmark was being >>>>messed up by buffering. >> >>>It's likely buffering. There's not a drive in existence that can >>>transfer at that rate without some kind of hardware or software >>>buffering. People like to point out the _potential_ speed of >>>SCSI hardware, but they all ignore the fact that no SCSI drive >>>even approaches this potential. The potential in an IDE might >>>be lower than a SCSI's potential, but so far they are pretty >>>equal in the real world. >> >>I wouldn't really say that... Quite a few of the high-end drives are >>actually rather decent. I have a reference for a MO disk system that >>sustains 12MBytes/second (I have no idea how much this costs...) >> >>Low-end SCSI's versus IDE might be about the same... > >I wonder how "local-bus" versus ISA bus fits into this comparison? >I haven't seen many local-bus SCSI controllers, but local-bus IDE >interface cards are becoming more common...so, what *is* the dominating >factor as far as data transfer rates: the disk drive, the >interface card/controller, the bus...? > Least we not forget, IDE drives are ISA specific. A 32 bit local bus IDE controller will not increase throughput with any significance due to the fact that the controller to drive interface is, in essence, a 16 bit ISA bus. Also, studies have shown that having local bus video AND local bus IDE in the same machine can actually degrade performance. ..... just the ramblings of an old fool ..... -- +-------------------------+-------------------------------------+ | Timothy E. Peoples | RazorByte Software Systems | | tep@rzrbyte.fay.ar.us | On The Cutting Edge of Technology | +-------------------------+-------------------------------------+ | Intelectual Philanthropy is our greatest gift; support GNU | +---------------------------------------------------------------+