At 08:57 PM 9/22/2001 -0400, Michael Maraist wrote:
> I don't remember if this has already been experimented with; I've
>lost track of what the previous sets of benchmarks were for. I was
>curious to see how much faster a single-function switch based interpreter
>would go.
Alas, whether a switch is faster or not seems to depend on the CPU. (And we
can't put all the functions in the switch, since we can redefine some of
the opcode functions) It's one of the reasons that we've got the
semi-abstract loop body, so we can use a switch or table dispatch as
appropriate for each platform.
We really ought to generate some benchmark numbers soon, though, for the
various methods on a variety of platforms.
(And, FWIW, a 2 or 3% speedup *is* worth the effort--imagine how many times
parrot will execute code, and then imagine the cumulative effects of that
2-3 percent...)
Dan
--------------------------------------"it's like this"-------------------
Dan Sugalski even samurai
[EMAIL PROTECTED] have teddy bears and even
teddy bears get drunk
