Skip to content
This repository

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Browse code

-README changes.

  • Loading branch information...
commit 23a7b05023670284fa9f2cb5083c23dd9071a451 1 parent 7622a73
Juan Manuel authored

Showing 1 changed file with 15 additions and 2 deletions. Show diff stats Hide diff stats

  1. +15 2 README.md
17 README.md
Source Rendered
@@ -48,7 +48,13 @@ Displayed times are in milliseconds.
48 48
49 49 Run avgtime without arguments to see more usage help.
50 50
51   -Don't forget to checkout the nice histogram:
  51 +Don't forget to checkout the *nice histogram*.
  52 +
  53 +*Note* that the following histogram shows two peaks because
  54 +of the CPU speed switching that laptops do. Some laptops
  55 +will often switch to low speed even in "High Performance" mode
  56 +and plugged to the wall. Forcing to high speed (always on)
  57 +eliminates the second peak (how to do that is beyond the scope).
52 58
53 59 avgtime -q -h -r1000 ls /etc -lR
54 60
@@ -91,6 +97,12 @@ Don't forget to checkout the nice histogram:
91 97 110: 1
92 98 111: 2
93 99
  100 +You can see the importance of the histogram now, and of
  101 +statistics such as the 'sample mode', 'median', and
  102 +standard deviation of the sample. All these together might
  103 +help *visualize and quantify* performance effects that you
  104 +wouldn't see by timing your program just once.
  105 +
94 106
95 107 How to install
96 108 ==============
@@ -143,7 +155,8 @@ version under test.**
143 155
144 156 See, one is not measuring the running time of a program. One is
145 157 actually measuring the interval of running times where it
146   -is likely to always stay... its probability distribution.
  158 +is likely to always stay... to know more about its probability
  159 +distribution.
147 160
148 161
149 162 _(Look for 'Hypothesis Testing' and 'Confidence Intervals' on the web

0 comments on commit 23a7b05

Please sign in to comment.
Something went wrong with that request. Please try again.