Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!rutgers!mtune!mtung!pgf
From: pgf@mtung.ATT.COM (Paul Fox)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Interleave factor on a 6300
Message-ID: <985@mtung.ATT.COM>
Date: Thu, 13-Aug-87 08:43:44 EDT
Article-I.D.: mtung.985
Posted: Thu Aug 13 08:43:44 1987
Date-Received: Sat, 15-Aug-87 08:37:23 EDT
References: <2928@mtgzz.UUCP> <1844@ttrdc.UUCP> <300@cpsc6b.cpsc6a.att.com>
Reply-To: pgf@mtung.UUCP (gws-Paul Fox)
Organization: AT&T ISL Middletown NJ USA
Lines: 24

In article <300@cpsc6b.cpsc6a.att.com> (C. R. Seaman) writes:
>No.  The interleave values for a hard disk are based on the performance of
>the disk, not the system.  It basically determines the number of cylinders
>the head must pass after reading from/writing to a cylinder before it can
>read from/write to another cylinder.  If the drive takes 4 cylinders to
>reset, and you set the interleave to 3, the head will wait for almost a
>complete revolution before attempting another read/write.  Setting the value
>to 5 would mean that the drive is basically idle for one cylinder pass.

    [ You should substitute "sector" for "cylinder" in the above,
	but that's not my main point. ]

	Are you sure about that?  I think you've just defined the minimum 
    interleave, all right, but don't forget the software overhead of setting 
    up for DMA transfers, managing buffers, application overhead in non-
    read-ahead systems, etc.  Those things all add to interleave needs as 
    well.  It's actually a very tough thing to get right on a given system --
    most UNIX boxes have low-level formatted interleaves, as well as an 
    interleave built into the ordering of the [initial] freelist... the two
    interleaves can actually work against one another if you're not careful.

-- 
			Paul Fox, AT&T Information Systems, Middletown NJ.
			  [ihnp4|vax135]!mtung!pgf (201)957-2698