Xref: utzoo comp.os.minix:6345 comp.sys.amiga:36900 comp.sys.amiga.tech:6175
Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!sun-barr!decwrl!shelby!polya!rokicki
From: rokicki@polya.Stanford.EDU (Tomas G. Rokicki)
Newsgroups: comp.os.minix,comp.sys.amiga,comp.sys.amiga.tech
Subject: Re: Minix for the Amiga ???
Message-ID: <10630@polya.Stanford.EDU>
Date: 15 Jul 89 19:51:40 GMT
References: <1610@uw-entropy.ms.washington.edu> <2882@ast.cs.vu.nl>
Sender: Tomas G. Rokicki <rokicki@Polya.Stanford.EDU>
Followup-To: comp.sys.amiga
Organization: Stanford University
Lines: 25

This is just a clarification for those who maybe don't know too much
about the Amiga, lest further misinformation spread . . .

> You forgot to ask the real question:  "How do you write an operating system
> for a computer that doesn't have a disk controller, but watches the bits
> come off the drive one at a time, in software?"

First of all, Andy's referring to the floppies only, not hard disks.
Secondly, the Amiga does indeed have a DMA channel associated with the
floppy drives, so the processor does not watch the bits come off the
drive one at a time; the processor goes off and works on another task.
On the other hand, the bits which are loaded by DMA are pretty raw,
consisting of clock and data bits interspersed.  Luckily, the blitter
can decode the MFM data quite rapidly.

The Amiga floppy drives are not slow.  The organization of the file
system makes listing directories slower than other operating systems,
but makes locating and opening files somewhat faster.  And were Andy
and his students to become familiar with the hard drive controllers
for the Amiga, they would be pleasantly surprised with the speed.

This is not a flame; I have a lot of respect for Andy Tanenbaum and
his work, of which Minix is a very fine example.

-tom