[net.micro] FIRST

gregr@tekig1.UUCP (02/28/84)

We all know that benchmarks are misleading and practically useless, often
making apples and oranges comparisons, but I simply couldn't resist one
Apple and IBM comparison.  The Byte Sieve of Eratosthenes benchmark is
particular poor but has become sort of a standard so I tried it out on
my IBM and a Macintosh.  Those of you that are above this sort of thing
or plan to just scoff should probably just ctrl-C out of this article
at this point.  I'll wait for you ..................


This is your last chance to keep from getting caught up in this 
foolishness ....ctrl-C......ctrl-C............

O.K. you were warned (couldn't resist could you!).

I've modified the original Byte Basic listing to speed up the performance
of the IBM (it should help any version of Microsoft Basic) and seems to
be fairer to Basic in comparison with other languages.  (Why doesn't he
just tell us the results??)  So that my test can be duplicated I'll list
the changes below.

	Byte version			My version

	120 while k <= 8190		120 if k > 8190 then 160
	130 	flags(k) = 0		130 for j=k to 8190 step prime
	140	k = k + prime		140 	flags(j) = 0
	150 wend			150 next

And now the results -

	IBM Basica		1197 secs
	Macintosh MS-Basic	 749 secs

I actually only ran a single iteration and multiplied by ten so that the 
times can be directly compared with oher results in Bytes survey.  
Notice how much faster the slight program change allows the IBM to run 
compared with Bytes results (1990 secs).  Seems fair to write the program 
the fastest way for a given language. 

One other simple benchmark - I tested the time to do 500 double precision
multiplies and divides.

	IBM Basica		12.7 secs
	Macintosh MS-Basic	 5.0 secs

Well, I told you this didn't mean anything so you wasted your own time
reading it.  But I wonder when the C compiler will be out???