Is Perl's unpack() ever faster than substr()?

Question

Several times I\'ve read that unpack() is faster than substr(), especially as the number of substrings increases. However, this benchmark suggests

Answer 1

I get similar results to the questioner under Ubuntu 9:

This is perl, v5.10.0 built for i486-linux-gnu-thread-multi
10
       Rate unpack substr
unpack 535925/s     --    -3%
substr 552749/s     3%     --
100
      Rate unpack substr
unpack 57957/s     --    -5%
substr 61264/s     6%     --
1000
     Rate unpack substr
unpack 4716/s     --   -22%
substr 6075/s    29%     --
10000
    Rate unpack substr
unpack 466/s     --   -24%
substr 609/s    31%     --
100000
     Rate unpack substr
unpack 46.3/s     --   -23%
substr 60.5/s    31%     --

But I'm not sure this is relevant. I don't tend to use unpack for simple string extractions, due to its unholy format string :-)

I think where it would shine is extracting encoded integers and all sorts of other binary information which is where I would use it.

One thing you should take from Matthew's and my (and your) benchmarks is that it will depend a lot on environmental factors. And keep in mind that speed, while good, is not the be-all and end-all - I don't think I've written much code that would be seriously affected by only being able to perform 4.6 million extractions per second rather than 6 million :-) You may need that extra performance but I doubt it for most applications written in Perl.