[Mono-list] esoteric programming
Tue, 28 May 2002 09:36:20 +0300
> Did you compile the sources with the Mono/MCS pair when comparing
> against MS? Currently MCS does not do two big groups of optimizations:
> loop inversion in for loops (trivial) and it uses more opcodes than
> necessary for tests/branches than it should (there is a bug report filed
> against this).
> Are the results the same when running the same compiled code (with CSC)
> in Mono vs MS.NET?
the code is not compiled with mcs or csc, it's compiled with the bf
the only thing that is compiled with mcs or csc is the bf compiler, but that
has no effect on what the bf compiler outputs.
> Also, are the times taken at function entry/leave or not?
> Because there might be a penalty in Mono currently, as we lack
there's only one function ;-)
the compiled program contains only about 10 to 15 of the simplest opcodes. I
don't think the problem is in the lack of 'ahead-of-time' compilation as the
program is very small and it has passed all possible code-paths within a
second. The real problem becomes apparent a few seconds later when it
becomes increasingly difficult to find the next prime number.