Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!oliveb!sun!burgundy!jborza
From: jborza%burgundy@Sun.COM (Jim_Borza)
Newsgroups: comp.sys.ibm.pc
Subject: Re: RLL drives and interleave factors
Keywords: RLL, hard disk drives, interleave, HDTST125, SPINTEST
Message-ID: <96540@sun.Eng.Sun.COM>
Date: 29 Mar 89 23:22:22 GMT
References: <1382@blake.acs.washington.edu>
Sender: news@sun.Eng.Sun.COM
Lines: 32

In article <1382@blake.acs.washington.edu>, djo7613@blake.acs.washington.edu (Dick O'Connor) writes:
> I'm trying to determine what interleave the Seagate ST238 30 MB RLL hard
> disk drive in my Zenith Z159 XT-compatible is set to.  
> [...] 
> I've downloaded two products from SIMTEL20 that do interleave testing.
> SPINTEST, which told me that my drive [...]
> was running at a 22:1 interleave (Eek!).
> [...] HDTST125, told me the interleave was 4:1 and that I
> could get slightly better performance by re-interleaving at 5:1.
> [...]
> Both packages seem RLL-aware.  Question is, who's right?  Maybe I need to
> find a *third* package, and go for 2 falls out of 3...
> 

SPINTEST probably reported it took 22 revolutions to read the disk - if true,
it only implies that the interleave is a tad too low.  Had it been MUCH too
low, it would have reported taking 26 revolutions (apparently, an occasional
"hit" is made).  I would accept the 4:1 reported by HDTST125 as accurate and
re-format with 5:1 interleave.  SPINTEST should then report taking 5 revs to
read the disk.  When using SPINTEST, the rule of thumb is: when the number
of revolutions is equal to the interleave, your disk subsystem is working
OK; if the number of revs is much higher than the interleave, the interleave
is too low.  The old-timers' method is to start with, say 6:1, run SPINTEST
and check if revs=interleave.  Decrease interleave by 1 and try again.  When
the revs jumps up, increase the interleave by 1 and leave it there.  BTW,
I include the interleave in the Volume label on the disk and also in a
comment in AUTOEXEC.BAT (I have several machines - all seem to require dif-
ferent i'leaves).  Volume label= yymmdd_i_ss (i=interleave, ss=DOS).


Jim Borza - Sun Microsystems
Disclaimer?  Sure, why not?