Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!usc!snorkelwacker!apple!amdahl!tetons!bdg
From: bdg@tetons.UUCP (Blaine Gaither)
Newsgroups: comp.arch
Subject: Re: 64 bits for times....
Message-ID: <8526@tetons.UUCP>
Date: 29 Aug 90 15:42:57 GMT
References: <703@exodus.Eng.Sun.COM> <67633@sgi.sgi.com> <957@halley.UUCP>
	<1990Aug24.181208.29581@rice.edu> <41090@mips.mips.COM> <AGLEW.90Aug25130251@dwarfs.crhc.uiuc.edu>
Organization: Amdahl Corp., Rexburg, ID
Lines: 14
In-reply-to: aglew@dwarfs.crhc.uiuc.edu's message of 25 Aug 90 17:02:51 GMT

I must agree with aglew.  You need high frequency timers (= cpu clock).

Even if you are timing small routines the extra precision is needed to
help you determine whether or not what you are observing is indeed
what you wish to observe.

I have seen countless situations where analysts have assumed some
small anomaly was "handling timer interrupts, .."  only to find out
it was a indication of a major source of error.

A final important reason for high precision timers is to help
architects manage the software implementation.  The cavalier way in
which OS types often treat timing facility implementation is only
exacerbated by letting them hide behind a coarse grain clock.