http://qs1969.pair.com?node_id=384967


in reply to •Re: Benchmark, -s versus schwartzian
in thread Benchmark, -s versus schwartzian

Aha! Moving the glob out of the loops (using Ambrus's code) gave the expected results:
/etc/ contains 311 files Benchmark: running Ordinary, Schwartzian, Strange for at least 5 CPU s +econds... Ordinary: 5 wallclock secs ( 1.47 usr + 4.09 sys = 5.56 CPU) @ 99 +.28/s (n=552) Schwartzian: 5 wallclock secs ( 3.95 usr + 1.13 sys = 5.08 CPU) @ 3 +33.46/s (n=1694) Strange: 5 wallclock secs ( 4.01 usr + 1.12 sys = 5.13 CPU) @ 35 +9.06/s (n=1842)

/bin/ contains 111 files Benchmark: running Ordinary, Schwartzian, Strange for at least 5 CPU s +econds... Ordinary: 5 wallclock secs ( 1.30 usr + 3.98 sys = 5.28 CPU) @ 34 +8.86/s (n=1842) Schwartzian: 6 wallclock secs ( 3.97 usr + 1.45 sys = 5.42 CPU) @ 9 +79.89/s (n=5311) Strange: 5 wallclock secs ( 3.84 usr + 1.46 sys = 5.30 CPU) @ 11 +45.85/s (n=6073)
One thing, If I used "/usr/bin" or any other path with a '/' in the middle it went haywire. Thanks a lot for all the help everybody!

Replies are listed 'Best First'.
Re^3: Benchmark, -s versus schwartzian
by ambrus (Abbot) on Aug 23, 2004 at 08:20 UTC

    This still does not explain the original results. I think that even with the glob in the loop, the Schwartzian transform should be faster. (I couldn't reproduce those results, the Schwartzian was faster for me even with your original code.)

    One thing, If I used "/usr/bin" or any other path with a '/' in the middle it went haywire.

    That's wierd.