Xref: utzoo rec.humor:18830 comp.misc:5078
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cwjcc!tut.cis.ohio-state.edu!rutgers!att!cuuxb!dlm
From: dlm@cuuxb.ATT.COM (Auntie Dion)
Newsgroups: rec.humor,comp.misc
Subject: Re: Looking for Computer Folklore
Keywords: benchmark
Message-ID: <2485@cuuxb.ATT.COM>
Date: 11 Feb 89 06:29:49 GMT
References: <6507@boulder.Colorado.EDU>
Reply-To: dlm@cuuxb.UUCP (Dennis L. Mumaugh)
Followup-To: rec.humor
Organization: ATT Data Systems Group, Lisle, Ill.
Lines: 26

In article <6507@boulder.Colorado.EDU> loughry@tramp.Colorado.EDU (J. Loughry) writes:
>
>(This is just a rumor, but it's a *neat* rumor....)
>
>	It seems (allegedly) that certain Microsoft compilers are smart
>	enough to figure out when they are being benchmarked.  Any time

I have heard of a certain compiler vendor that designed their floating
point subroutines for optimum use with bench marks.  Certain routines
cache their last argument and its value, for example:
	int getpid(){
		static int pid = 0;
	
		if( pid	) return pid;
		return( pid = _getpid());
	}
	where _getpid is the UNIX system call.

This is also done for certain well known arguments in trig functions.
What with optimizers inserting inline function code and moving
invariant function calls outside of a loop, its a wonder a benchmark
means anything these days.

-- 
=Dennis L. Mumaugh
 Lisle, IL       ...!{att,lll-crg}!cuuxb!dlm  OR cuuxb!dlm@arpa.att.com