Path: utzoo!attcan!uunet!wuarchive!udel!princeton!gauss!markv
From: markv@gauss.Princeton.EDU (Mark VandeWettering)
Newsgroups: comp.arch
Subject: Re: Bloat costs
Message-ID: <1020@idunno.Princeton.EDU>
Date: 3 Jul 90 14:58:41 GMT
References: <1990Jun27.011149.2406@Stardent.COM> <63034@sgi.sgi.com> <9570@celit.fps.com> <2273@l.cc.purdue.edu>
Sender: news@idunno.Princeton.EDU
Reply-To: markv@gauss.Princeton.EDU (Mark VandeWettering)
Distribution: na
Organization: Princeton University
Lines: 19

In article <2273@l.cc.purdue.edu> cik@l.cc.purdue.edu (Herman Rubin) writes:

>When can people in the computer field realize that this will always be the
>case?  There is much that a compiler cannot handle well, and this will 
>remain the case.  Computers cannot beat a human adept at something as
>simple as chess, let alone the ingenuity required to produce efficient
>computational procedures.

*sigh*  Computers can certainly beat ME at chess, and I would imagine 
that most good compilers can beat most average assembly hackers on 
a given problem.  In particular, humans can be bad at effeciently managing
registers (let's just save 'em all).  Compilers ARE getting better, and its
not clear that humans are, so draw your own conclusion.

Also, once you have a good compiler, you needn't incur the cost of training
a person to become a proficient assembly language programmer.  (Therapy 
is expensive! :-)

Mark VandeWettering