Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!mit-eddie!uw-beaver!tektronix!tekcrl!tekgvs!toma
From: toma@tekgvs.TEK.COM (Thomas Almy)
Newsgroups: comp.lang.c,comp.sys.m68k
Subject: Re: Ackermann's function is wrong in Moto Benchmark Report
Message-ID: <2309@tekgvs.TEK.COM>
Date: Thu, 21-May-87 11:10:52 EDT
Article-I.D.: tekgvs.2309
Posted: Thu May 21 11:10:52 1987
Date-Received: Sat, 23-May-87 14:44:30 EDT
References: <16731@amdcad.AMD.COM>
Reply-To: toma@tekgvs.UUCP (Thomas Almy)
Organization: Tektronix, Inc., Beaverton, OR.
Lines: 29
Xref: mnetor comp.lang.c:2224 comp.sys.m68k:496

In article <16731@amdcad.AMD.COM> tim@amdcad.AMD.COM (Tim Olson) writes:
>
>In the "Apples to Oranges" MC68020 Benchmark Report Motorola put out in
>1986, the C code for Ackermann's function was shown as:
[ some code deleted ]

>int x,y;
>{
>	if (x==0) return (++y);
>	if (y==0) return (A(--x,1));
>	return (A(--x,A(x,--y)));
>}
>The last line is incorrect. [ ... ] 
>The last line should be written:
>
>	return (A(x-1,A(x,--y)));


Not only that, but using all of those decrement operations turns the
benchmark into a good compiler test (rather than a processor test).
Is the C compiler smart enough to realize that storing the incremented y
(in the first case) or decremented x or y (latter cases) is not necessary?
I wouldn't be happy unless the benchmark was:

	if (x==0) return (y+1);
	if (y==0) return (A(x-1,1));
	return(A(x-1,A(x,y-1)));

Tom Almy