# from Sisyphus # on Friday 19 January 2007 02:39 am: >Well ... I can't detect a significant difference for a 10,000,000 > element array of integers.
Surprisingly, the ref version is slower than the list version if you put it inside a bit of a loop. That's actually counter to the pure-perl behavior. I suppose the XS code is basically accessing the array ref already. I stuck in a sum just to make sure there isn't too much optimizer magic happening. Anyhow, yeah. Use the list. I think the subroutine call overhead shows up more dramatically if you move zeros from $elements to $times. http://scratchcomputing.com/tmp/bm_ref_vs_list.pl Benchmark: timing 5 iterations of by_list, by_ref, perl_list, perl_ref... by_list: 4 wallclock secs (...) @ 1.47/s (n=5) by_ref: 4 wallclock secs (...) @ 1.35/s (n=5) perl_list: 28 wallclock secs (...) @ 0.18/s (n=5) perl_ref: 25 wallclock secs (...) @ 0.20/s (n=5) --Eric -- The only thing that could save UNIX at this late date would be a new $30 shareware version that runs on an unexpanded Commodore 64. --Don Lancaster (1991) --------------------------------------------------- http://scratchcomputing.com ---------------------------------------------------
