1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
|
#!/usr/bin/env python
import sys
from scipy.stats import mannwhitneyu
SIGNIFICANCE_THRESHOLD = 0.0001
a,b = {},{}
for (path, d) in [(sys.argv[1], a), (sys.argv[2], b)]:
for line in open(path):
try:
tokens = line.split()
samples = tokens[:-1]
label = tokens[-1]
d[label] = map(float, samples)
except:
pass
common = set(a.keys()).intersection(b.keys())
ps = []
for key in common:
_, p = mannwhitneyu(a[key], b[key]) # Non-parametric t-test. Doesn't assume normal dist.
am, bm = min(a[key]), min(b[key])
ps.append((bm/am, p, key, am, bm))
ps.sort(reverse=True)
def humanize(ns):
for threshold, suffix in [(1e9, 's'), (1e6, 'ms'), (1e3, 'us'), (1e0, 'ns')]:
if ns > threshold:
return "%.3g%s" % (ns/threshold, suffix)
maxlen = max(map(len, common))
# We print only signficant changes in benchmark timing distribution.
bonferroni = SIGNIFICANCE_THRESHOLD / len(ps) # Adjust for the fact we've run multiple tests.
for ratio, p, key, am, bm in ps:
if p < bonferroni:
str_ratio = ('%.2gx' if ratio < 1 else '%.3gx') % ratio
print '%*s\t%6s -> %6s\t%s' % (maxlen, key, humanize(am), humanize(bm), str_ratio)
|
