Path: utzoo!attcan!uunet!husc6!bloom-beacon!tut.cis.ohio-state.edu!mailrus!ames!amdahl!pyramid!prls!mips!mash
From: mash@mips.COM (John Mashey)
Newsgroups: comp.arch
Subject: Re: RISC machines and scoreboarding
Message-ID: <2477@winchester.mips.COM>
Date: 30 Jun 88 08:31:43 GMT
References: <1082@nud.UUCP> <2438@winchester.mips.COM> <1098@nud.UUCP> <2465@winchester.mips.COM> <318@mucmot.UUCP>
Reply-To: mash@winchester.UUCP (John Mashey)
Organization: MIPS Computer Systems, Sunnyvale, CA
Lines: 36

In article <318@mucmot.UUCP> ron@mucmot.UUCP (Ron Voss) writes:
>In article <2465@winchester.mips.COM>, mash@mips.COM (John Mashey) writes:
>> 	c) Figuring that a RISC compiler should do optimization, but worrying
>> 	that this feature might be buggy, is like worrying about the safety
>> 	of flying in a 747 and bringing your own seat-belt because you
>> 	don't really trust Boeing to remember to include them :-)

>I have observed a few cases where hours were wasted searching for a
>programming bug, only to find that the CISC (68020) compiler's optimizer
>is buggy (so one turns on the optimizer only for a release, since it
>slows compilations anyway, thoughouly test again after turning on, etc.).

>Are there really *good* reasons to put more trust in RISC optimizers?

>Methinks the metaphor a bit overdone.

Sorry, let me try again.  I wasn't trying to say optimizers were easy
or safe at all: they're not, in fact, they are like building a 747.
The point was, in fact, that if you're doing an optimizer (building
a 747), THAT's what you worry about, because it's hard.  You don't
worry about getting nops after loads ("this feature") (seatbelts),
because if a somebody can't get THAT right, they have no business
building optimizers (747s).  [No disrespect to Boeing, I just picked the
biggest plane I've flown in.]

1) I have marginally more trust in RISC optimizers than in CISC ones,
on general principles [i.e., at least some parts are simpler.]

2) What I mostly trust/distrust is the optimizer's author(s);
I'd much rather have an expert's CISC optimizer than a turkey's RISC one.
(Fortunately, I get to use a super-expert's RISC optimizer. :-)
-- 
-john mashey	DISCLAIMER: 
UUCP: 	{ames,decwrl,prls,pyramid}!mips!mash  OR  mash@mips.com
DDD:  	408-991-0253 or 408-720-1700, x253
USPS: 	MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086