Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!mips!cs.uoregon.edu!ogicse!intelhf!ichips!iwarp.intel.com!inews!pima!bhoughto
From: bhoughto@pima.intel.com (Blair P. Houghton)
Newsgroups: comp.lang.c
Subject: Re: low level optimization
Message-ID: <4063@inews.intel.com>
Date: 27 Apr 91 03:29:55 GMT
References: <21818@lanl.gov> <3905@inews.intel.com> <20270@alice.att.com>
Sender: news@inews.intel.com
Organization: Intel Corp, Chandler, AZ
Lines: 25

In article <20270@alice.att.com> wilber@homxb.att.com writes:
>One can of course come up with various schemes involving macros or comments

Or #pragma's...

>that inform your favorite compiler that the parameters aren't aliased, but the
>problem is that there are no standards for how this is to be done, so any
>such information is likely to be understood *only* by your favorite compiler,
>and no other.

It's likely to be so, anyway.  If I have invented an optimization
that makes 6x gains in speed of operations on pointer-accessed
data, how fast do you expect me to propagate that to my marketplace
competitors?  Patent Office or no, it's going to take a long time
and cost someone a lot of money to get this little feature into
any more compilers.

>Right now, in the Real World (TM), optimized Fortran really does run faster
>than optimized C.  A situation that truly sucks.

Funny, that happened in '64, too.

				--Blair
				  "...everyone's figuring
				   out how old we were..."