[lmi] [PATCH] Show standard deviation and maximum duration in timing tests

[Vadim Zeitlin]

 Hello,

 Here is the promised simple patch adding the display of the standard
deviation and maximum, in addition to the mean and minimum, to
AliquotTimer:

        https://github.com/vadz/lmi/pull/150/

 The last comment in this PR shows how the tests look right now, but I'll
reproduce it here for convenience:

---------------------------------- >8 --------------------------------------
Running calendar_date_test:
  Speed tests...
  Aggregate    : 3.866e-06 s avg,     1 us SD       4 min      13 max (2587 
runs)
  Construct    : 1.681e-07 s avg,     0 us SD       0 min       9 max (59506 
runs)
  Assign       : 2.763e-08 s avg,     0 us SD       0 min       9 max (361974 
runs)
  Stepping     : 4.165e-08 s avg,     0 us SD       0 min      11 max (240089 
runs)
  Get y, m, d  : 2.920e-08 s avg,     0 us SD       0 min       9 max (342411 
runs)
  Format       : 4.357e-07 s avg,     0 us SD       0 min      10 max (22954 
runs)
  Calculate age: 1.128e-07 s avg,     0 us SD       0 min       9 max (88676 
runs)
  DOB limit    : 3.451e-06 s avg,     1 us SD       3 min      14 max (2898 
runs)

  Speed tests for production code...
       10 bytes: 6.900e-05 s avg,     3 us SD      66 min      92 max (100 runs)
      100 bytes: 7.056e-05 s avg,     5 us SD      68 min      89 max (100 runs)
     1000 bytes: 7.568e-05 s avg,     3 us SD      74 min      95 max (100 runs)
    10000 bytes: 1.837e-04 s avg,    20 us SD     171 min     296 max ( 55 runs)
   100000 bytes: 1.152e-03 s avg,   139 us SD    1056 min    1509 max (  9 runs)
  1000000 bytes: 9.112e-03 s avg,     9 us SD    9103 min    9120 max (  2 runs)
 10000000 bytes: 9.831e-02 s: first trial took longer than 1.000e-02 s desired 
limit
---------------------------------- >8 --------------------------------------

 This is not perfect, but is easily readable, thanks to vertical alignment,
and is still almost short enough to fit into 80 columns, so it seems like
the best compromise. But please let me know if you prefer something else.

 Thanks in advance,
VZ

pgpPOk7QKmh8A.pgp
Description: PGP signature