Path: utzoo!attcan!uunet!mcvax!unido!estevax!wck353
From: wck353@estevax.UUCP (HrDr Weicker Reinhold )
Newsgroups: comp.arch
Subject: Stoned benchmarks and other naming conventions
Keywords: benchmark, Dhrystone, Whetstone
Message-ID: <470@estevax.UUCP>
Date: 23 Feb 89 14:20:13 GMT
Organization: Siemens AG ESTE3 Erlangen, W-Germany
Lines: 45


In article <5681@pdn.nm.paradyne.com>, Alan Lovejoy
talks about the name ?h*stone for benchmarks.
There are similar tendencies in other parts of the
computer world:
- CPU architecture names have to be four-to-five-letter words (originally:
  four-letter :-), with an I and an S in the middle (*IS*).
- Operating system names have to be four-to-five-letter words (originally:
  four letter), with the last two letters IX (*IX).
- (Integer) performance numbers have to be called "MIPS" - whatever
  the definition might be (native MIPS, peak MIPS, sustained MIPS, RISC MIPS,
  VAX MIPS, EDN MIPS, nonsense MIPS ...).
- Benchmark program names can be arbitrarily long, but they should end
  with "stone", and the second letter must be an "h" (?h*stone).
  If people insist on the more narrow tradition, the name should somehow
  indicate a degree of humidity.

Since I started the last tradition with the 1984 CACM "Dhrystone"
paper (at least, I don't know of any other "*stone" except
Whetstone prior to Dhrystone), I probably should acknowledge to whom I
owed the name: I had written an internal predecessor program that was used
in the years 1980-83 by Siemens and Intel. When I saw that people appreciated
having such a benchmark program, I made a revised version and submitted
it for publication. Justin Rattner from Intel suggested the names "Dryestone"
or "Dhrystone"; I choose the latter one.

Not everyone knows the origins of "Whetstone", so I may mention it here:
The original publication (Curnow/Wichmann: A Synthetic Benchmark;
The Computer Journal 19 [1976], 1, 43-49) does not contain the word
"Whetstone", but the statistical data this program was derived from
had been collected with the Whetstone ALGOL compiler system.
As far as I remember (I don't recall a written source), the National
Physical Laboratory, where this work was done, is located in a
town or village called "Whetstone". Does anyone in netland know
whether this is true? And when did the word "Whetstone" become a
nickname of this benchmark?

By the way, I like the title of a paper at the Summer USENIX Conference 1987:
"Taking Performance Evaluation out of the 'Stone' Age" (K.J. McDonell).

-- 
Reinhold P. Weicker, Siemens AG, E STE 35, PO Box 3220, D-8520 Erlangen, Germany
Phone:		     +49-9131-720330 (Centr.Europ.Time, 8 am - 5 pm)
UUCP:		     ...!mcvax!unido!estevax!weicker
Disclaimer:	     Although I work for Siemens, I speak here only for myself