From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=0.7 required=5.0 tests=BAYES_00,INVALID_DATE,
	MSGID_SHORT,REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no
	version=3.4.4
Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!cs.utexas.edu!uunet!philmtl!pedersen
From: pedersen@philmtl.philips.ca (Paul Pedersen)
Newsgroups: comp.lang.ada
Subject: Re: Inheritance & limited private types
Message-ID: <800@philmtl.philips.ca>
Date: 27 Oct 89 15:44:28 GMT
References: <48695@ricerca.UUCP> <6845@hubcap.clemson.edu> <126675@sun.Eng.Sun.COM> <31018@news.Think.COM>
Reply-To: pedersen@philmtl.philips.ca (Paul Pedersen)
Organization: Philips Electronics Ltd. -  St. Laurent  P.Q., Canada
List-Id: <comp.lang.ada>

In article <31018@news.Think.COM> barmar@kulla (Barry Margolin) writes:
>
>Making a program efficient without a profiler is like making a program work
>without a debugger.  Sure, you can manually sprinkle your code with print
>statements and timing statements, but it's not an efficient way to do it.
>

I spent some time last year trying to profile a large UNIX-based system
running on a 386.  Turned out to be a waste of time since profiling is
driven by UNIX clock-ticks which, at least on our system, is 100t/sec.

I found that the software (under test) could execute up to 30 function
calls between ticks, so most functions never got hit by the interupt, and
all time-spent-in-function's were meaningless.

Speeding up the interrupt frequency is apparently not a good idea since
most time will then be spent in context-switch's.

The conclusion I came to was that it is not possible to do profiling 
using UNIX on a fast processor, short of using a logic analyzer and a
lot of interpreting software (far from sure that this is even feasable).

I would be *very* interested in hearing from anybody who has solved this
problem (I gave up) :-)

Paul Pedersen (pedersen@philmtl)