Path: utzoo!utgpu!watmath!clyde!mcdchg!chinet!bigtex!texbell!killer!chasm
From: chasm@killer.DALLAS.TX.US (Charles Marslett)
Newsgroups: comp.sys.ibm.pc
Subject: Re: RLL Hard Disks
Summary: RLL and bus speed are independent problems
Message-ID: <5951@killer.DALLAS.TX.US>
Date: 30 Oct 88 03:23:57 GMT
References: <1541@valhalla.ee.rochester.edu> <584@tank.uchicago.edu>
Organization: The Unix(R) Connection, Dallas, Texas
Lines: 34

In article <584@tank.uchicago.edu>, ads4@tank.uchicago.edu (adam david sah) writes:
> RLL IS unreliable now. Now only because the new generation of machines has 
> higher clock speeds, which exceed the capability of RLL controllers to 
> necessarily encode the data fast enough. One friend of mine had to slow 
> down his AT from 12mhz to 6mhz in order to ensure 100% writing! The results 
> previously were disastrous. If anyone can confirm this, please reply publicly!
> 
> -A.Sah'88

The problem your friend had is probably due to the controller card not working
properly with the bus speed (as you mentioned), but that has very little to do
with the drive data encoding algorithms.  His best bet to find a controller
that works well at the full bus speed is to get one reccommended by the
motherboard manufacturer or to try a brand new this year super duper bang up
expensive controller (I don't know what the current one is . . . (;-).

Seriously, RLL controllers are almost always faster than the MFM ones (and
slower than the EDSI drives most of the time, too).  They are more often
buffered which also speeds them up, and they are newer technology also speeding
them up.  As a result, they are often configured to run with reduced wait states
and this will lead to problems if the motherboard is running significantly
faster than an 8 MHz AT -- Everex Step machines are especially bad (good?) in
this respect because they also run a funny symmetry clock as well.  A
possibility is that the RLL controller can be reconfigured for standard
numbers of wait states in this case.

In conclusion, RLL controllers are about as likely as any others to fail
in any particular machine -- but the box maker or the RLL card maker
should know enough to offer some reccommendations and/or offer some
kind of fix.

Charles Marslett
STB Systems, Inc.        <-- apply all standard disclaimers, we are not
chasm@killer.dallas.tx.us    in this business yet!