Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!mit-eddie!ll-xn!ames!elroy!mahendo!jplgodo!wlbr!etn-rad!jru
From: jru@etn-rad.UUCP (John Unekis)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Intel Microprocessors
Message-ID: <265@etn-rad.UUCP>
Date: Mon, 31-Aug-87 12:19:39 EDT
Article-I.D.: etn-rad.265
Posted: Mon Aug 31 12:19:39 1987
Date-Received: Fri, 4-Sep-87 07:12:16 EDT
References: <1112@lznv.ATT.COM> <399@aucs.UUCP> <3225@cucca.columbia.edu>
Reply-To: jru@etn-rad.UUCP (0000-John Unekis)
Organization: Eaton Inc. IMSD, Westlake Village, CA
Lines: 40

In article <7144@steinmetz.steinmetz.UUCP> davidsen@crdos1.UUCP (bill davidsen) writes:
>In article <257@etn-rad.UUCP> jru@etn-rad.UUCP (0000-John Unekis) writes:
>|In article <7042@steinmetz.steinmetz.UUCP> davidsen@crdos1.UUCP (bill davidsen) writes:
>|>....
>Obviously you would rather slander BYTE than admit that the 80386 was
>faster in these benchmarks. Moreover you were so eager to disagree with
>the results that you totally missed the point, which is the validity of
>the benchmarks, 
....... 
No ,I did not miss the point, you deleted from my posting the results
that I posted which showed the times that I got when I reran the BYTE
benchmarks under controlled conditions.  My objection to the BYTE benchmarks
was that they did not measure the actual CPU usage, they used a stop watch
to measure the time between printed messages from their test program. The
80386 benchmarks were run under a special 32 bit protected mode software 
environment which I believe interfered with the printing of these messages
and thus invalidated the timing of the test. 

I reran the BYTE fibonacci sequence test under controlled conditions, and
instead of the 20 to 1 advantage that BYTE claimed for the 80386, an
accurately measured(using UNIX time command)test showed that a 16Mhz 80386 was
only 12% faster than a 12Mhz 68020. 

I therefore have posted the results of this counter-benchmark in the hope
of showing how badly inaccurate benchmarks can distort performance figures.

I routinely use several CPU's in my work, including the 80386, 8088, 80286,
68020, 32032 and some lesser-known processors. None of them has any 
advantages that would give them a greater than 1 order of magnitude speed
advantage in real life situations. I prefer the 68020 to the intel family
beacuase it does not require me to deal with segmentation.

A little light-hearted hand-to-hand verbal combat in defense of one's 
favorite CPU is all in good fun. But I refuse to sit still when grossly
inaccurate tests are presented as gospel truth.

---------------------------------------------------------------------

disclaimer: You can't blame anyone but me for what I do.
ihnp4!wlbr!etn-rad!jru