NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/misc-part2/12430

Internet Message Format | 1996-08-05 | 3.7 KB

Path: EU.net!sun4nl!xs4all!usenet
From: [e-mail] (Jeroen T. Vermeulen)
Newsgroups: comp.sys.amiga.advocacy,comp.sys.amiga.misc
Subject: Re: Why are europeans dumb enough to buy amigas?
Date: Sun, 14 Apr 96 16:42:53
Organization: Leiden University, Mathematics & Computer Science, The Netherlands
Message-ID: <[e-mail]>
References: <[e-mail]> <[e-mail]> <[e-mail]>
NNTP-Posting-Host: asd01-15.dial.xs4all.nl
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
X-NewsSoftware: GRn 2.1 Feb 19, 1994


In article <[e-mail]> [e-mail] (David Corn) writes:
> On Tue, 9 Apr 1996 12:30:04 GMT, [e-mail] (Finn Higgins)
> wrote:

> >yep.  but you don't get all that much extra functionality out of it when you
> >register.  Plus, I'd like to see you emulate a Mac at the same speed on your
> >PC for the same amount of money.
>
> Granted that Emulator costs money and lacks a few features (quite a
> few, granted!).  However, it is a nice solution for those with a very
> fast pc - even on slower PCs like the 486100, 040/25 or 33 speeds are
> attainable.  That's not at all bad.

In fact it's questionable.  UAE doesn't reach the speed of a plain 7 MHz 68000
on a Pentium, according to the author.

For any interpreter, and there's less difference with dynamic translation than
emulator writers would like, you'd need at least three data memory accesses just
to be able emulate an 040 instruction.  That alone would account for several 486
cycles per emulated instruction (at least three if the D cache is
single-ported), but in half the L1 cache size (because 040 I fetches effectively
become 486 D loads).  On top of that come the register spills of the emulator
itself.

Then come Little-Endian conversion, status bits, and exceptions.  That's worth
at least a cycle per instruction as well.  Finally of course you've got to
perform the actual instruction.  So even assuming plain 68000 emulation, no MMU
and no FPU (which is much slower on the 486 if you ask me), you'd need to create
some kind of handicap to slow down the 040 or the 486 won't keep up with only
three times the clock speed (and about the same bus speed).  Without a L2 cache
the 486 doesn't stand a chance at all.

Of course the 040's more orthogonal design does make it a lot easier to emulate:
No scores of undocumented instructions, no Appendix H crap, no messy segmented
memory model, no self-modifying code.

Wait a minute!  No self-modifying code?  You're not talking about a 68k system
that does this, are you?  Because that would mean you'd either have to disable
the caches or flush them all the time, which would bias the comparison in favour
of an emulator.


>                                             Now, how fast can an Amiga
> emulate a PC?  And what are the relative percentages of users that see
> a need to emulate another platform, on both platforms?

Don't really know, but I do know that I used to compile my Pascal assignments
with a PC compiler on my A500.  Worked like a charm, except for the lousy "OS"
(or really just "program loader") I had to run on the PC side.  It's a funny
feeling to see a machine thinking in 16-bit numbers as late as the 1980's.  Some
say they're still doing it now...

--
============================================================================
#  Jeroen T. Vermeulen   \"How are we doing kid?"/   Yes, we use Amigas.   #
#---  [e-mail]    ---\"Oh, same as always."/--         ...          --#
#[e-mail] \ "That bad, huh?"  /  Got a problem with that? #
Moderator for alt.support.cluelessness