Path: utzoo!mnetor!uunet!husc6!ut-sally!ut-emx!auscso!bigtex!james
From: james@bigtex.uu.net (James Van Artsdalen)
Newsgroups: comp.unix.xenix
Subject: Re: a reply to ISC on 'clarifications'
Message-ID: <783@bigtex.uu.net>
Date: 3 Feb 88 20:22:28 GMT
References: <795@vixie.UUCP>
Organization: F.B.N. Software, Austin TX
Lines: 67
Summary: ESDIs & drive types

In article <795@vixie.UUCP>, paul@vixie.UUCP (Paul Vixie Esq) writes:
> [...]  First, no BIOS drive-type table I've ever
> seen, nor any that my motherboard vendors and manufacturers have ever seen
> or heard of, has even one single entry in it for a 34-sector device.  Perhaps
> Compaq's machines are the exception, but in general, 17-sector entries are
> the only thing you'll get in your average 386 BIOS.

PC's Ltd supports 34 sector tracks in BIOS in several variants (I can't get to
the ROM-based setup program from unix to see exactly how many).  But this
shouldn't be necessary.  I am told that with the Western Digital boards you
set the drive type to #1, and the board makes adjustments from there.  The
person I talked to who's done it doesn't know how the controller does this.
The OMTI 8620 is apparently set to drive type #0 (??? so I am told - the
controller looks to see what's out there, if anything).

> [ general discussion of WD1005 and lack of BIOS 18 head support ... ]
> Note that the 1005-WAH has no support for
> floppy drives, so you'll need a WD-1002-FOX, which fits in a short slot.
> Note also that WD Tech Support told me specifically on the phone that the
> controller was only useful at 3:1 interleave.

Careful when ordering the WD1002/FOX: lots of places will ship you an XT
hard disk controller: apparently there is another WD1002/??? that isn't the
same as the /FOX card.

Since the WD1005 has no track buffering, I would not expect it to be capable
of 1:1 interleave.  Perhaps if the peripheral runs at 12MHz you might be able
to do 2:1 interleave: I don't know where the cutoff point is.  Another argument
for the WD1007.

One warning about the WD1007: it does not work well with more than 35 sectors
per track, and even at 35 sectors per track you can't use 'em all.  At 35
sectors per track the hard disk will use the 35th to replace any bad sectors
in the first 34, giving you a full 34 per track, but you can't use all 35
per track directly.  I know of at least one Micropolis and one CDC drive (ESDI)
that support as many as 37 sectors per track via jumpers.  I don't really know
what to make of this: perhaps someone else can explain.  I am wary of trying
to use 37 though if the manufacturer only ships the drive set at 34...

> [ discussion of a neat Adaptec 2322 board ... ]
> So far,
> though, I say: stay away from the 1005-WAH, look into the Adaptec 2320/2322.

The WD1007 is a better board than the WD1005 anyway.  But neither can handle
ST506 drives and neither have floppy drive support.  There is an OMTI 8620
controller which handles any combination of ESDI and ST506 drives (both of
either type or a mixture), and handles 1.2meg and 360K all in one.  Like the
Adaptec above, this one is also 1:1 interleave with buffering and automatic
configuration at power-up.  This card cost me $175, and is also currently on
order.  I will be setting it up with a CDC Wren III-182, an ST-4096 and a
1.2meg floppy on a PC's Ltd 386/16 (running uPort SV/386 L2.2).  Send mail if
you'd like to know how it turns out.

There is another OMTI, the 8627, that drops ST506 in favor of RLL, but is
otherwise the same.  Use RLL at your own risk.

One caveat: I've heard rumors that the buffering hard disk controllers don't
do much for Xenix.  I've heard this from a couple of places, but not from
anyone with firsthand experience.  Has anyone actually tried both a buffering
and nonbuffering controller on the same hardware/software Xenix machine to see
if there is a major difference?  One of the 386 machine manufacturers has had
several reports that the WD1007 is no better than a WD1005 with Xenix, and
there is a comment along this line also in the 386 mailing list.  Does Xenix do
track-buffering in the kernel?
-- 
James R. Van Artsdalen    ...!uunet!utastro!bigtex!james     "Live Free or Die"
Work: 512-328-0282 Home: 346-2444; 110 Wild Basin Rd. Ste #230, Austin TX 78746