In case you haven't read them already, the usual performance tips are here:

http://julia.readthedocs.org/en/latest/manual/performance-tips/

The first and foremost being that using non-constant globals is a
significant trap.

On Mon, Mar 23, 2015 at 3:17 PM, Michael Bullman <[email protected]>
wrote:

> Hi All,
>
> So I recently translated a program from Python using PyPy to julia. I was
> hoping to see some speed improvements with the move, but instead I have a
> seen a fairly significant speed decrease. Which I'm thinking is more user
> error than language differences.
>
> for the python program I'm able to call it directly as "pypy
> load_balancer_sim.py" and it will run automatically and complete.
>
> With Julia, since I developed it using the suggested Module
> method/procedure I open julia then load the module, then call the relevant
> method. All from the REPL, then the program runs. I'm noticing this takes
> significantly more time to run. I'm thinking two things could be having a
> major impact.
>
> 1) it's still inside the module wrapper
> This should be a relatively easy fix since I should just have to remove
> the top and bottom lines of code.
>
> 2) I'm calling it from the REPL.
> Initially I didn't think this would have much of a difference, but I'm
> wondering if this introduces unnecessary overhead.
>
> Then just sorta related to this general question of calling a script from
> the command line directly, does Julia have a way to run a main method? In
> python I know you can use syntax similar to
>
> "if __name == '__main__':
>    code code code
> "
>
> And the script will automatically run that bit of code when called. Is
> there a similar syntax in Julia?
>
> Thank you all for any help
>

Reply via email to