Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uwm.edu!csd4.csd.uwm.edu!gblock
From: gblock@csd4.csd.uwm.edu (Gregory R Block)
Newsgroups: comp.sys.amiga.advocacy
Subject: Re: Amiga basher
Message-ID: <13192@uwm.edu>
Date: 17 Jun 91 22:10:46 GMT
References: <1991Jun17.160936.14622@cs.mcgill.ca>
Sender: news@uwm.edu
Reply-To: gblock@csd4.csd.uwm.edu
Lines: 102
Originator: gblock@csd4.csd.uwm.edu

From article <1991Jun17.160936.14622@cs.mcgill.ca>, by genius@cs.mcgill.ca (Michel NGUYEN):
> 
> It's clearer to me now. Any process at -1 will have maximum priority
> over the other ones at > -1.

No, it just got worse.  :)

Because the Mandel process is running at -1 (probably the lowest
task), it catches everything that isn't being used by anything else,
therefore, if something else needs time, it is given it immediately,
and the Mandel process is held.  But since nothing in the foreground
is processor intensive (due to the fact that the Amiga _NEVER_ uses
polled IO), even a 2400 baud download takes little processor time, and
almost no slowdown exists.  I often raytrace in the background at -2
(I have other things at -1).  Now, if the raytrace is all I'm doing,
and I set it at 5, it takes 30 minutes.  When I drop it into the
back at -2 when I run it, it takes 35 doing a 60 minute ZModem
download.  (just so the transfer was continuous, and that the cpu was
ALWAYS loaded down just that much).  Not bad, really.  :)

> That's however will not change anything at all. Even if you have the best
> real-time process scheduling in the world, and if the CPU is already
> running at 100%, when you throw in a CPU intensive apps, the overall

Right, but it takes quite a bit to get that cpu to run at 100%.  QUITE
a bit.  Something like a raytrace, or a monstrous spreadsheet, or
perhaps a mandelbrot generator.  And then if you give it a priority of
one below the lowest task (they're usually all at 0), it will eat all
unnecessary cpu cycles that would usually go unused.

> performance will be degraded, not matter what. But I guess it does not
> apply in your case as most of the time you seem to have a lot 
> (if not only) of *interactive* apps with iddle and waiting time.

Yes, I do use a lot of interactive apps.  I also use a lot of
interactive ones, such as The Art Department Professional, Imagine
(3d raytracing/anim/objed), and various other thingies.  :)

> You cannot throw in an intensive CPU apps into a CPU already working at 100%
> and expect the same overall performance. If you still don't understand what
> I mean, tell me and I'll mail you an uudecoded TIFF picture.

So you run the intensive CPU stuff at a lower priority than the
interactive stuff, and since the majority of interactive stuff is
waiting, and since there is no such thing as busy waiting, 99.9% of
the CPU time goes to your intensive apps.

> I know you can play a couple of animated demos, play a stereo sound music,
> downloading a couple of meg of files and archiving and compute a Mandelbrot
> image (320x200x32??) and format a diskette with your Amiga very nicely. But
> I don't. I don't format diskettes everyday, I don't play demos everyday,
> I rarely download files at 2400 baud. But I work everyday. So while I work, 
> I work. Sure I play games, but I don't swicth between work and 
> games/demo/sounds as easyly as the Amiga or Unix do. 
> Maybe it's my flaws.

So what DO you do?  And if you had real multitasking, you can always
find use for it.  As a current IIcx owner, I can vouch.

I'll admit, it's not a real world example above, but okay, it does
show you what can be done.

> Yes it takes a "17 MIPS" machine for X-windows, but it is more than just 
> any window manager or process manager or GUI, it has all the system protocol 
> for communication and networking (Client/server), that's where the real
> power of X and it's not an OS. But I guess you already know that also.
> What are the flaws in X? Have you ever try X in Amiga 3000UX? Windows 
> in a 386sx is faster. 

Have you ever tried X under AmigaDOS?  It's faster than my IIcx, and
it's on a 2000...

> Throw in Mathematica, a couple of rendering or fractal computations and
> you will see how poor is ANY design. When you will need more than 
> 17 MIPS (but I doubt it considering what you do most of the time) one day,
> I guess you will not upgrade to a 68040, it only shows how poor design
> the 68030 is.

Yes, but imagine how much better if the Mac's OS were as efficient as
AmigaOS, and offered better multitasking than it currently does?  It's
the reason that a NeXT runs an 030 at 5 mips, and an Amiga runs it at
7.  Everything's the same, but the OS is different.  Architecture is
also something to consider...  But it does at least hint at what I'm
trying to say.  And I DO consider the NeXT environment to be better
than 7.0...  Not as nice as AmigaOS, but it's Unix.  As a Unix
workstation, it's a lovely os.  :)

> Yeah, but it took more than 7 mega years for EVOLUTION to get from
> a simple Ameoba to Homo Sapiens, because of backwards compatibility.

Ooh, I didn't know I was compatible with an Amoeba.  :)

That could lead to some really strange college parties.  :D

Greg


-- 
Socrates:  "I drank WHAT????"
LMFAP:  "Next time you see me, it won't be me."
Wubba:  "A dream is nothing more than a wish dipped in chocolate and sprinkled
with a little imagination." (From my poem, "A Dream")			-Wubba