On 07/09/2010 02:23 PM, David Mertens wrote:
On Fri, Jul 9, 2010 at 11:24 AM, Judd Taylor
<[email protected] <mailto:[email protected]>> wrote:
The truth is that doing anything on the extremes requires a programmer
be clever. PDL just allows you to do this better than Perl.
Even the might Perl TIMTOWTDI runs out of ideas when it comes to
computing huge datasets like this. At the very minimum, PDL allowore
alternatives at being clever to make these things work quickly.
For me personally, if I want something very simple that is
guaranteed to
be very fast, I go immediately to PDL::PP and write a quick sub to do
this. You can spend 5 minutes using Inline PDL::PP to develop what
works, and then put the code in a library somewhere for future
use. You
get code that's easier to maintain that way, IMO, as it doesn't
need to
be as "clever" as the perl level PDL code.
-Judd
Writing a quick sub using PDL::PP is fantastically simple - if you
know what you're doing. This is why I wanted to have my first talk be
on PDL::PP, not PLplot. But the masses spoke, and I focused on
plotting instead.
I work with datasets with thousands of elements, and I recently
replaced two perl nested for-loops with an Inline::Pdlpp sub. The
compute time went from a few seconds to seemingly instantaneous
feedback. Unfortunately, there's no genlte introduction to using PDL::PP.
there is Inline::Pdlpp, which for some reason comes up as PP-Inline.html
in the html docs (maybe so it doesn't collide with PDL::PP), and thus
probably doesn't show up in your PDL::Index page, but it's there.
http://pdl.perl.org/?docs=PP-Inline&title=Inline::Pdlpp
maybe you've seen this, maybe you haven't.
Derek
David
--
Sent via my carrier pigeon.
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
