[Octave-bug-tracker] [bug #51630] BIST test for log2 fails on Windows

[Markus Mützel]

Follow-up Comment #11, bug #51630 (project octave):

I am getting strange readings with your test script on Windows.
Running with hg id e56d84a16a2e (frexp from the standard lib), I see:

>> bm_log2
N =  50
mean_bm =  0.22193
median_bm =  0.22228
std_bm =  0.0029001

>> bm_log2
N =  50
mean_bm =  0.22318
median_bm =  0.22218
std_bm =  0.0031263

>> bm_log2
N =  50
mean_bm =  0.22213
median_bm =  0.22172
std_bm =  0.0022295

>> bm_log2
N =  50
mean_bm =  0.22010
median_bm =  0.21983
std_bm =  0.0019340


With hg id 9401e88f63cf (gnulib):

>> bm_log2
N =  50
ans =  0.20613
ans =  0.20505
ans =  0.0075337

>> bm_log2
N =  200
ans =  0.20497
ans =  0.20462
ans =  0.0019133

>> bm_log2
N =  50
mean_bm =  0.20438
median_bm =  0.20409
std_bm =  0.0015047


So it executes actually about 8% faster with gnulib than with the standard
library function (unless something else overcompensated the slowdown between
these changesets). Both examples run with the same vector of 10e6 random
doubles.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?51630>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/