Speed-up for loops

[David Cournapeau]

On Sun, Sep 5, 2010 at 8:28 PM, BartC <bartc at freeuk.com> wrote:

>
> One order of magnitude (say 10-20x slower) wouldn't be so bad. That's what
> you might expect for a dynamically typed, interpreted language.

10/20x slower than C is only reached by extremely well optimized
dynamic languages. It would be a tremendous achievement. If that's
what you are after, look at LUA with its JIT, or scheme + stalin.

For cases where vectorization is indeed not applicable (recursive
algorithms), like in some signal processing, there are specialized
tools which are very expressive while retaining good speed (faust is
an interesting one for audio signal processing).

> That would simply be delegating Python to a scripting language.

That's a strange thing to say if you compare it to matlab.

> It would be
> nice if you could directly code low-level algorithms in it without relying
> on accelerators

It would be nice, but the fact is that python cannot do it  - and is
quite far from being able to do it. I don't think it is as important
as you think it is, because things like numpy are extremely powerful
in many cases.

cheers,

David