Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!biar!jhood
From: jhood@biar.UUCP (John Hood)
Newsgroups: comp.unix.xenix
Subject: Re: ESDI vs SCSI
Message-ID: <467@biar.UUCP>
Date: 14 Apr 89 16:19:47 GMT
References: <6628@mrspoc.UUCP> <3400006@cpe>
Reply-To: jhood@biar.UUCP (John Hood)
Organization: Biar Games Inc., Ithaca, NY
Lines: 31

In article <3400006@cpe> neese@cpe.UUCP writes:
>The one thing that many people overlook in a disk implementation is the
>fact that while the kernel is in the driver code, everthing else is dead.
>The kernel cannot do any scheduling for user processes.  So in an ST506/ESDI
>driver, the kernel is dead, until the driver gets the block(s) that the
>kernel requested.

This doesn't seem correct to me-- I think you mean to say that the
kernel is dead *while the driver is actually reading the block(s) from
the WD-style controller*.  AT ST-506 style controllers are perfectly
capable of interrupting into a device driver when a requested sector
becomes available after after seek, rotational, and read delays.
The drivers don't sit there polling the controller until the sector
comes ready; they yield to other processes in the meantime.  The
Adaptec SCSI adapter still holds a lead theoretically, since the
driver is only running at the start and completion of I/O, whereas the
ST-506 style controller interrupts for every physical sector.

Frankly, I don't see much difference between the Tandy 4000 with
ST-506 drives and Microport at home and the Tandy 4000 with the 80 meg
SCSI subsystem and Tandy-oid Xenix at work.  (True, it's a comparison
with a few too many variables in it.)  I'll allow that the difference
might be greater under a multitasking, multiuser sort of load, but
from a single-user perspective, there's little difference between the
two.

  --jh-- 
John Hood, Biar Games snail: 10 Spruce Lane, Ithaca NY 14850 BBS: 607 257 3423
domain: jhood@biar.uucp (we hope) bang: anywhere!uunet!biar!jhood
[food for disclaimer readers]
[special dessert tidbit for broken mailers]