Path: utzoo!attcan!uunet!lll-winken!ames!mailrus!um-math!hyc
From: hyc@math.lsa.umich.edu (Howard Chu)
Newsgroups: comp.sys.atari.st
Subject: Re: Atari-bashing (was Re: Looking for an Evangelist)
Message-ID: <540@stag.math.lsa.umich.edu>
Date: 9 Jan 89 07:30:39 GMT
References: <474@ur-cc.UUCP> <6847@spool.cs.wisc.edu> <12835@cup.portal.com> <428@ho7cad.ATT.COM> <3216@sugar.uu.net> <538@stag.math.lsa.umich.edu> <84396@sun.uucp>
Sender: usenet@math.lsa.umich.edu
Reply-To: hyc@math.lsa.umich.edu (Howard Chu)
Organization: University of Michigan Math Dept., Ann Arbor
Lines: 48
UUCP-Path: {mailrus,umix}!um-math!hyc

In article <84396@sun.uucp> cmcmanis@sun.UUCP (Chuck McManis) writes:
>> Sure, it irks me that I can't pop up Uniterm while I'm compiling
>> something. Big deal, it doesn't take very long to compile stuff anyway.
>> Fast filesystems are nice that way. That's also an interesting point
>> you fail to summarize in your table - the ST and PC disk I/O are
>> probably the fastest around. 
>
>And this is where is all breaks down. That second sentence is called a 
>rationalization. When I program I often start up a paint program to design
>icons, start other compiles, and pass data back and forth between my compiler
>envrionment and editing environment. Multitasking is a lever that is quite
>powerful, if it weren't Bill Gates and John Scully wouldn't be trying so
>damned hard to get it on their systems no? Finally, if you get a chance

True 'nuff. I guess I shouldn't be harping on this point, particularly
since I use Minix anyway. (Hey, anyone interested in a port of TOS to a
Sun 3-280? Just kidding, really. I think...  }-)

>try the Amiga Fastfilesystem on a Quantum 80S. Try on a freshly formatted
>one and one that is 80% full. We'll even let you use 16Meg partitions
>and the Amiga will smoke your eyes. We're talking 800 - 900K bytes/sec
>transfer rates bud.

Well, we're talking pretty nice, I guess. I've got a Quantum 80S on my
machine. I've compared it to an Amiga with the same drive. (At the local
computer store, of course.) Copying 1 megabyte files around on a freshly
formatted (yes, 16 MB, oh well...) partition. My ST was faster. (Did I
say "was?" Still is, too, I think...) Given a sequential read, figure that
having to read 1 meg ought to be enough to defeat any speedups due to
track caching and such. Though since they were two identical drives, you
wouldn't expect any of that to be an issue at all. The Amiga guy at the
store suggested that the ST gained from a 1:1 sector interleave. Since the
ST DMA controller is supposed to max out at 1MB/sec, you can see that the
difference wasn't all that great. But, we clocked the Amiga around 500-600K
per second, and the ST at around 700-800K.

I suppose this can't really be all that conclusive. After all, it's a
mechanical device, prone to inconsistencies from imperfections. (I mean,
a couple times the ST clocked at 1 MB per second. Many times it timed in
a lot slower, like 500K or so. What can you tell from that? Seek delays?
That should only be a couple milliseconds difference, right?)

--
  /
 /_ , ,_.                      Howard Chu
/ /(_/(__                University of Michigan
    /           Computing Center          College of LS&A
   '              Unix Project          Information Systems