On Tue 08 Mar 2011 00:10, Michael Ellis <[email protected]> writes:

> Thanks for the education, Andy, and congrats again on the guile
> compiler implementation.  Your version handily outperforms the fastest
> python implementation I know.

Thanks, but there is still a ways to go.  If guile can count to 10
million in half a second, that means it's spending 100+ cycles per loop,
when it could be spending 5 or something.  I figure we have a good
factor of 4 to squeeze out easily with native compilation, and more with
flow analysis and untagging and such.

Regards,

Andy
-- 
http://wingolog.org/

Reply via email to