Beefy Boxes and Bandwidth Generously Provided by pair Networks
Clear questions and runnable code
get the best and fastest answer

Re^2: bit array comparison

by Amendil (Novice)
on Oct 22, 2019 at 17:46 UTC ( #11107865=note: print w/replies, xml ) Need Help??

in reply to Re: bit array comparison
in thread bit array comparison

Very interesting!

To be honest, I'm not really interested in space.
The tsv I'm playing with has about 500k lines, and I can already run my algorithm with everything in ram.

It's just that my naive implementation is slow, and I want to see how fast it can be. I know that I lost a lot of performance when I started comparing list of strings.
I just expect bit array to be much much faster where applicable. To confirm that, I'll be able to compare with other similar columns that cannot be turned into bit array.

Thank you for making me learn a bit more about Perl today.


To reproduce what you did for Jaccard module, calling set_to_string first, I also did the same for your implementation.
I makes sense in my actual context as I first read the whole file, cleaning rows, etc. so creating the bit array will be part of the process.

What's interesting is that it doubled the performance hash: 340k/s -> 420k/s, bits: 800k/s -> 1.7m/s
Also, using `my $ic = unpack '%b*', pack 'N', $int;` provides slightly better performance (+1% consistently) that your implementation with Hamming weighs.
Now, I need to read a bit about Hamming weighs. Working with bits is always interesting, but it feels kind of a cipher.

Replies are listed 'Best First'.
Re^3: bit array comparison
by rjt (Curate) on Oct 23, 2019 at 01:16 UTC

    Oh yes, there's still more room for optimization if you need it. If you can't get the speed you need out of pure Perl, you can also use Inline::C or XS for maximum speed. If the Pure perl solution is good enough though, I'd stick with that. I love benchmarking; results still find a way to surprise me.

Log In?

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://11107865]
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others browsing the Monastery: (3)
As of 2021-04-12 02:49 GMT
Find Nodes?
    Voting Booth?

    No recent polls found