Path: utzoo!mnetor!uunet!nuchat!peter
From: peter@nuchat.UUCP (Peter da Silva)
Newsgroups: comp.sys.amiga
Subject: Re: AmigaDos don't thrash no more!
Message-ID: <665@nuchat.UUCP>
Date: 20 Feb 88 19:22:12 GMT
References: <8504@sunybcs.UUCP>
Organization: Public Access - Houston, Tx
Lines: 52

In article <8504@sunybcs.UUCP>, ugpete@sunybcs (Peter Theobald) writes:
> AmigaDos should sort disk requests by track...
> 	This would eliminate thrashing, and would speed up disk accesses
> to boot! I think this is similar to what a clone of Peter da Silva meant by
> single-threading loadSegs.

Alas, this wouldn't help. Here's what would happen.

Proc a: loading a file that's on tracks 10 and 11.
Proc b: ditto for 20 and 21
Proc c: ditto for 30 and 31

All procs do their first requests.

Proc a gets its block back, and requests the next block.
Meanwhile disk is heading out to satisfy procs b and c.
Proc b gets its block back, and requests the next block.
Meanwhile disk is heading out to satisfy c.
Proc c gets its block back, and requests the next block.
Meanwhile disk is heading back to satisfy procs a and b.
Proc b gets its block back, and requests the next block.
Meanwhile disk is heading back to satisfy proc a.
Proc a gets its block back, and requests the next block.
Meanwhile disk is heading out to satisfy procs b and c.

The problem is that ALL requests for ALL files aren't out there at the same
time.

What would help would be a few track buffers. Executable files tend to be
pretty contiguous, so:

All procs do their first requests.

Proc a gets its block back, and requests the next block.
Meanwhile disk is heading out to satisfy procs b and c.
However the next 10 requests from proc A will be satisfied from the track
	cache.
Proc a finally requests a block on another track...
Proc b gets its block back, and requests the next block.
Meanwhile disk is heading out to satisfy c.
However the next 10 requests from proc B will be satisfied from the track
        cache.
Proc c gets its block back, and requests the next block.
Meanwhile disk is heading back to satisfy procs a and b.
However the next 10 requests from proc C will be satisfied from the track
	cache...

Still shuffling, but 11 times as fast.
-- 
-- a clone of Peter (have you hugged your wolf today) da Silva  `-_-'
-- normally  ...!hoptoad!academ!uhnix1!sugar!peter                U
-- Disclaimer: These aren't mere opinions... these are *values*.