[Shootout-list] ackermann, values for ''n''
Bengt Kleberg
bengt.kleberg@ericsson.com
Tue, 07 Jun 2005 14:57:10 +0200
greetings,
first i would like to thank Mr Brent Fulgham for indulging me by
answering some of my emails about adding more, and more spread out,
''n'' to the tests. i should work harder at the problem, but since i am
almost the only person interested in the problem i sometimes laps into
prolonged periods of inactivity.
it seems to be difficult to get the kind of ''n'' selection i would like
to see, given the current 24 hour limit for running all tests on all
languages. i would like to suggest that we look at the possiblity to
have another limit. what about 1 hour for all the tests for one
language? or 1 hour for one test for all the languages? or something?
i have looked at the ''n'' values for ackermann. only beacuse it is the
first function in the list. however, it is a rather good starting point
since it is relly difficult to handle with my usual method. ackermann
goes from very fast to very slow with a very small increase in ''n''.
the current value (9) gives run times from 0.1 second to (almost) 100
seconds. 0.1 is too short (imho). if i increase ''n'' to 10 i get
sligthly below 0.5 seconds, still a bit short. with 11 i get 3 seconds.
this is enough (0.5 to 1 second would be ok, i think), but 11 should
make the 100 second program approach 3000 seconds.
i think it would be better to have another test than ackermann. one that
is not so sensitive to ''n''.
ok?
bengt