Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site terak.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!hao!noao!terak!doug
From: doug@terak.UUCP (Doug Pardee)
Newsgroups: net.works
Subject: Re: Ackermania
Message-ID: <547@terak.UUCP>
Date: Mon, 13-May-85 14:52:35 EDT
Article-I.D.: terak.547
Posted: Mon May 13 14:52:35 1985
Date-Received: Thu, 16-May-85 05:54:44 EDT
References: <1863@topaz.ARPA>
Organization: Terak Corporation, Scottsdale, AZ, USA
Lines: 31

I probably should've let this rest, but I can't resist...

When I selected the Ackermann function to "benchmark" HLLs vs. assembler
I did so knowing that because it was written recursively, the calls
would eat C alive.  I even recall saying so in my original posting.

But now I see news postings and get mail insisting that it was "unfair"
to multiply this effect by comparing a VAX against a Z-80A, because
the VAX is crippled by having an excruciatingly slow "call" instruction.
Now, I personally think that save-everything-in-the-machine-because-we-
don't-know-what-the-subroutine's-gonna-do type "call" instructions
are a direct result of HLL-mania and thus are fair game, but I'll give
the VAX the benefit of the doubt.

Dan Hoey was kind enough to pass along the non-recursive version of
Ackermann that he'd developed, so I benchmarked that version.  The
Z-80A didn't do nearly as well this time.

This time, it beat the VAX 750, but not the 780.

The time for 1 million computations of acker(3,6) (seconds):
  VAX 750  optimized C  -- 59.2u
  4MHz Z-80A  assembler -- 47.1
  VAX 780  optimized C  -- 34.4u

So I guess that for code which doesn't do a lot of "call" instructions,
you *can* make up the CPU time overhead of C code by trading in your
Timex/Sinclair 1000 for a VAX, as long as you get at least a 780.
-- 
Doug Pardee -- Terak Corp. -- !{ihnp4,seismo,decvax}!noao!terak!doug
               ^^^^^--- soon to be CalComp