Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!bloom-beacon!oberon!cit-vax!ucla-cs!zen!ucbvax!jade!eris!mwm
From: mwm@eris.BERKELEY.EDU (Mike (My watch has windows) Meyer)
Newsgroups: comp.sys.amiga
Subject: Re: AllocMem() with 16-bit Manx
Message-ID: <5210@jade.BERKELEY.EDU>
Date: Fri, 25-Sep-87 01:38:46 EDT
Article-I.D.: jade.5210
Posted: Fri Sep 25 01:38:46 1987
Date-Received: Sat, 26-Sep-87 18:40:14 EDT
References: <8709141841.AA06918@cory.Berkeley.EDU> <1584@gryphon.CTS.COM> <174@toylnd.UUCP>
Sender: usenet@jade.BERKELEY.EDU
Reply-To: mwm@eris.BERKELEY.EDU (Mike (My watch has windows) Meyer)
Organization: Missionaria Phonibalonica
Lines: 37

In article <174@toylnd.UUCP> dca@toylnd.UUCP (David C. Albrecht) writes:
<> >	I would suggest Aztec-C mainly because it will compile int's as either
<> >16 bits or 32 bits (Most Atari compilers use 16 bit ints), and you don't want
<> >to have to worry about a program compilable under BOTH 16 and 32 bit ints.
<> >
<There is a difference between saying 16 bit 'ints' and 16 bit 'shorts'.  As
<far as I know of Lattice 3.10 doesn't support 16 bit 'ints'.

Correct. 4.0 should have real 16-bit shorts. So we can all enjoy the
benefits of an OS that expects integers on the stack that are longer
than an 'int'.

<In any case, when it comes to using 16 bit ints in Lattice though I have
<to differ, they don't handle them very nicely.

Dig out the compiler, and start going through the output when you
declare things as shorts. You'll see lots of "EXT.W" and "EXT.L"
instructions. This follows C semantics, where all integer arithmetic
is done in the int type - or longer.

[Description of changing ints to shorts in Lattice with no gain, but
getting a benefit on the AT&T 3b1 (aka the Unix PC).]

Hmm. I wouldn't expect much of a gain in Lattice. It tends to generate
good code (more on that later, when I have the time). If the 3b1
compiler is as abysmal as most PCC's I've seen, then it might make a
lot of difference. Have you looked at the code coming out of Lattice
in both cases? And from the 3b1? Then again, the 3b1 compiler could
have the semantics wrong, and not being doing the extend to 32 bits.
Lattice 3.03 had that bug.

	<mike
--
I'm gonna lasso you with my rubberband lazer,		Mike Meyer
Pull you closer to me, and look right to the moon.	mwm@berkeley.edu
Ride side by side when worlds collide,			ucbvax!mwm
And slip into the Martian tide.				mwm@ucbjade.BITNET