•Re^3: Why does File::Find chdir?

That's not very deep, to go to /usr/bin/foo instead of using chdir first. What I meant was a deep hierarchy, like /usr/local/lib/X11/app_defaults/foo. Every path that has all those steps has to be looked up step-by-step, repeating the same work for the kernel over and over.

Admittedly, the cost is fairly cheap these days, since modern kernels can cache a lot of the intermediate directories. But it's still a non-zero cost, and while that might not make a difference for a dozen lookups, it will for a thousand lookups.

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.

Comment on •Re^3: Why does File::Find chdir?

Replies are listed 'Best First'.
Re^4: Why does File::Find chdir? by hossman (Prior) on Jul 14, 2004 at 06:06 UTC
Ah, yeah .. good point. For those who want to prove this to themselves, here's the script I used to convince myself... Read more... (2 kB) And the result... laptop:~/tmp/monk-test> ~/monk.pl -50 laptop:~/tmp/monk-test> ~/monk.pl 100 Benchmark: timing 100 iterations of chdir, no_chdir... chdir: 2 wallclock secs ( 1.39 usr 0.40 sys + 0.12 cusr 0.10 +csys = 2.01 CPU) @ 55.87/s (n=100) no_chdir: 3 wallclock secs ( 1.50 usr 1.08 sys + 0.09 cusr 0.17 +csys = 2.84 CPU) @ 38.76/s (n=100) Rate no_chdir chdir no_chdir 38.8/s -- -31% chdir 55.9/s 44% -- laptop:~/tmp/monk-test> ls 0 1 a aa laptop:~/tmp/monk-test> rm -rf * laptop:~/tmp/monk-test> ~/monk.pl -100 laptop:~/tmp/monk-test> ~/monk.pl 100 Benchmark: timing 100 iterations of chdir, no_chdir... chdir: 4 wallclock secs ( 2.98 usr 0.71 sys + 0.13 cusr 0.12 +csys = 3.94 CPU) @ 27.10/s (n=100) no_chdir: 8 wallclock secs ( 3.03 usr 3.98 sys + 0.13 cusr 0.21 +csys = 7.35 CPU) @ 14.27/s (n=100) Rate no_chdir chdir no_chdir 14.3/s -- -47% chdir 27.1/s 90% -- [download]	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^4: Why does File::Find chdir?
by hossman (Prior) on Jul 14, 2004 at 06:06 UTC

Ah, yeah .. good point.

For those who want to prove this to themselves, here's the script I used to convince myself...

Read more... (2 kB)

And the result...

laptop:~/tmp/monk-test> ~/monk.pl -50
laptop:~/tmp/monk-test> ~/monk.pl 100
Benchmark: timing 100 iterations of chdir, no_chdir...
     chdir:  2 wallclock secs ( 1.39 usr  0.40 sys +  0.12 cusr  0.10 
+csys =  2.01 CPU) @ 55.87/s (n=100)
  no_chdir:  3 wallclock secs ( 1.50 usr  1.08 sys +  0.09 cusr  0.17 
+csys =  2.84 CPU) @ 38.76/s (n=100)
           Rate no_chdir    chdir
no_chdir 38.8/s       --     -31%
chdir    55.9/s      44%       --
laptop:~/tmp/monk-test> ls
0  1  a  aa
laptop:~/tmp/monk-test> rm -rf *
laptop:~/tmp/monk-test> ~/monk.pl -100
laptop:~/tmp/monk-test> ~/monk.pl 100
Benchmark: timing 100 iterations of chdir, no_chdir...
     chdir:  4 wallclock secs ( 2.98 usr  0.71 sys +  0.13 cusr  0.12 
+csys =  3.94 CPU) @ 27.10/s (n=100)
  no_chdir:  8 wallclock secs ( 3.03 usr  3.98 sys +  0.13 cusr  0.21 
+csys =  7.35 CPU) @ 14.27/s (n=100)
           Rate no_chdir    chdir
no_chdir 14.3/s       --     -47%
chdir    27.1/s      90%       --
[download]

[reply]
[d/l]
[select]