[Shootout-list] Ackermann
Einar Karttunen
ekarttun@cs.helsinki.fi
Sat, 18 Jun 2005 11:18:10 +0300
John Skaller <skaller@users.sourceforge.net> writes:
> Real partial evaluators don't *allow* you to
> distinguish which phase code runs in.
Then how about taking both N and M from the command line?
That should solve things if we want to play that game.
This would mean rewriting all the existing solutions
but should only take a few minutes for each one.
I don't think we have yet a such partial evaluator
in the shootout - or at least it is not winning.
Marking the C++ solution as an alternative and
coming back (with N from command line), when
we get a such partial evaluator sounds like
the best way.
> Of course, since we have partial evaluators,
> we cannot just measure run time. We MUST measure
> the whole time taken from source to result,
> as if the program were interpreted: compile time
> must be included.
This would be very nice for Forth ;)
In reality this would make the benchmarks
have little sense, as the ordering would
change very sharply with N (from
interpreters to compilers winning).
- Einar Karttunen