Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!rutgers!mtune!whuts!homxb!homxc!rps
From: rps@homxc.UUCP (R.SHARPLES)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Formatting a Hard Disk (changing interleave, response)
Message-ID: <796@homxc.UUCP>
Date: Wed, 29-Jul-87 15:47:14 EDT
Article-I.D.: homxc.796
Posted: Wed Jul 29 15:47:14 1987
Date-Received: Sat, 1-Aug-87 01:47:07 EDT
References: <1604@bellcore.bellcore.com> <553@cup.portal.com>
Distribution: world
Organization: AT&T Bell Laboratories, Holmdel
Lines: 46
Summary: What interleave does

In article <553@cup.portal.com>, Dion_L_Johnson@cup.portal.com.UUCP writes:
> I recall some results (from a source I have forgotten) that found
> that changing the interleave away from 1:1 made very little
> difference in performance.  Anyone know more?  Is there a reason
> besides performance why we would want to change the interleave?

The interleave is the measure of whether consecutive data is stored in
consecutive sectors.  1:1 interleave means consecutive data is stored
in physically consecutive sectors.  If the head is finished
reading sector 1 (example) but the controller has not passed all the data on
to the CPU yet, it must wait before reading sector 2 (S2) (the next sector with 
data for 1:1 interleave).  Since S2 is physically right after S1, the 
head will have to wait a full drive revolution before S2 comes around 
again.  However if the interleave was 1:2 (or 2:1?) the data would be
recorded in every other sector: S1-S3-S5-etc..  Then the head will
read S1 and while S2 is passing underneath the controller will finish
passing S1 data.  Just as the controller is done S3 will start to pass
under the head and it will be read on schedule.

An interleave of 1:3 means that every third sector is read consecutively,
1:4 every forth.  The average Seagate 225 with WD controller uses
and interleave of 1:3.  1:4 is recommended on 8086 machines for some 
reason.

If interleave is too SMALL, eg. less than 1:3, the head will always have to
wait for the drive to make a full revolution between reading 
consecutive sectors and hence will slow the data transfer rate.
For example: an ST225 w/WD controller with 1:3 interleave on a PC6300 
will have a transfer rate of about 25Kb/sec.  With 1:4 the transfer
rate will rise to about 160Kb/sec.  That is the difference between
catching the next sector and waiting a full revolution.

You won't really notice this unless you are moving a lot of data onto
and off of the disk.  To load an 80K program will take 3 seconds with
the 1:3 interleave or .5 seconds with the 1:4 interleave.  It is
a personal choice whether this difference matters.


Russ Sharples
homxc!rps

NOTE:

The above in NO WAY reflects the opinions of AT&T.
These opinions are my own and the results of un-scientific and 
highly irregular analysis methods.