Beefy Boxes and Bandwidth Generously Provided by pair Networks
Perl-Sensitive Sunglasses
 
PerlMonks  

Re^5: Performance problem with Clone Method

by dwm042 (Priest)
on Jul 28, 2011 at 03:33 UTC ( [id://917174] : note . print w/replies, xml ) Need Help??


in reply to Re^4: Performance problem with Clone Method
in thread Performance problem with Clone Method

I read this thread through 3-4 times before I "got it", which is that PDL is really a floating point engine that makes things like singular value decomposition a one liner.

And that back in 1983, the kind of programming the OP is interested in would have been written in assembler and/or C, and so bitwise efficiency is to be optimized here.

D-

Replies are listed 'Best First'.
Re^6: Performance problem with Clone Method
by etj (Deacon) on May 26, 2022 at 14:01 UTC
    PDL is really a floating point engine that makes things like singular value decomposition a one liner
    This is not and never was correct. PDL has always been a numerical processing library (or engine, if you prefer), which has both bindings to various external libraries which can do linear algebra in a highly-optimised way, or read images, or search for Ngrams in DNA, etc, and also includes built-in operations for floating-point linear-algebra stuff like SVD, and operations on integers. This includes bitwise operations on integers, which is nearly the most efficient way to do boolean stuff (the most efficient way would be to treat each of the application's data items as a bit, and compact those 8 per byte, but your code has to do that itself, at least for now).