Yes, this is one of the most surprising things about Julia. Perhaps the time 
has come to put a warning about this right at the start of a REPL session.

 — John

On Nov 22, 2014, at 4:50 AM, Viral Shah <[email protected]> wrote:

> Try putting everything in a function.
> 
> http://julia.readthedocs.org/en/latest/manual/performance-tips/
> 
> -viral
> 
> On Saturday, November 22, 2014 6:16:39 PM UTC+5:30, Ronald L. Rivest wrote:
> I'm a julia newbie trying to get a sense of how efficient the Julia compiler
> is, for a very simple little program.  I attach copies of
> foo.py and foo.jl, which compute the same thing.
> 
> I then ran them (for the python version with both
> pypy and with python).  The results:
> 
>     pypy:         2 seconds
>     julia:        30 seconds
>     python:    43 seconds
> 
> I was surprised to see julia so badly beaten by pypy,
> and to see that the difference between julia and ordinary
> python was less than a factor of two... 
> 
> Is there an optimization trick or optimization switch that
> I missed for Julia??
> 
> Thanks!
> 
> Cheers,
> Ron Rivest
> 
> 11:36:12 temp $ date; pypy foo.py; date
> Fri Nov 21 23:36:22 EST 2014
> 0.726947459371
> Done.
> Fri Nov 21 23:36:24 EST 2014
> 11:36:24 temp $ date; julia foo.jl; date
> Fri Nov 21 23:36:37 EST 2014
> 0.726947459370689
> Done.Fri Nov 21 23:37:07 EST 2014
> 11:37:07 temp $ date; python foo.py; date
> Fri Nov 21 23:37:32 EST 2014
> 0.726947459371
> Done.
> Fri Nov 21 23:38:15 EST 2014

Reply via email to