forked from spcl/npbench
-
Notifications
You must be signed in to change notification settings - Fork 0
/
plot_lines.py
130 lines (114 loc) · 4.5 KB
/
plot_lines.py
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
import math
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from npbench.infrastructure import utilities as util
# create a database connection
database = r"npbench.db"
conn = util.create_connection(database)
data = pd.read_sql_query("SELECT * FROM lcounts", conn)
# get rid of kind and dwarf, we don't use them
data = data.drop(['timestamp', 'kind', 'dwarf', 'version'],
axis=1).reset_index(drop=True)
# Remove everything that does not have a domain
data = data[data["domain"] != ""]
# for each framework and benchmark, choose only the best details,mode (based on min npdiff), then get rid of those
aggdata = data.groupby(
["benchmark", "domain", "framework", "mode", "details", "count"],
dropna=False).agg({
"npdiff": np.min
}).reset_index()
best = aggdata.sort_values("npdiff").groupby(
["benchmark", "domain", "framework", "mode"],
dropna=False).first().reset_index()
best = best.drop(['domain', 'mode', 'details'], axis=1).reset_index(drop=True)
frmwrks = list(best['framework'].unique())
print(frmwrks)
assert ('numpy' in frmwrks)
frmwrks.remove('numpy')
frmwrks.append('numpy')
percs = ["{}_perc".format(f) for f in frmwrks]
# get improvement over numpy (keep times in best_wide_time for numpy column), reorder columns
data = best.pivot_table(index=["benchmark"],
columns="framework",
values="count").reset_index() # pivot to wide form
data = data[['benchmark'] + frmwrks].reset_index(drop=True)
# get improvement over numpy (keep times in best_wide_time for numpy column), reorder columns
diffs = best.pivot_table(index=["benchmark"],
columns="framework",
values="npdiff").reset_index() # pivot to wide form
diffs = diffs[frmwrks].reset_index(drop=True)
for f in frmwrks:
data["{}_perc".format(f)] = (diffs[f] / data['numpy']) * 100
# color of the heatmap is percentage changed
colors = data[percs]
# rename the columns, drop the " Perc" for labelling
colors = colors.rename(columns={a: b for a, b in zip(percs, frmwrks)})
# number in the heatmap is change to NumPy (except for NumPy, where it is the total)
numbers = data[frmwrks]
for f in frmwrks:
if f == 'numpy':
continue
numbers[f] = numbers[f] - numbers['numpy']
plt.style.use('classic')
figsz = (len(frmwrks) + 1, 12)
fig, ax0 = plt.subplots(nrows=1, ncols=1, figsize=figsz)
# plot benchmark heatmap
im = ax0.imshow(colors.to_numpy(),
cmap='RdYlGn_r',
interpolation='nearest',
aspect="auto",
vmin=0,
vmax=100)
for i in range(len(data['benchmark'])):
for j in range(len(colors.columns)):
l = numbers.to_numpy()[i, j]
lo = l
p = colors.to_numpy()[i, j]
if not math.isnan(p):
p = str(int(p))
if j < len(colors.columns) - 1:
if math.isnan(l):
text = ax0.text(j,
i,
"missing",
ha="center",
va="center",
color="red",
fontsize=7)
elif l >= 0:
l = "+" + str(int(l))
else:
l = str(int(l)) #+ ", " + str(p) + "%"
if not math.isnan(lo):
text = ax0.text(j,
i,
l,
ha="center",
va="center",
color="white",
fontsize=10)
else:
if not math.isnan(lo):
text = ax0.text(j,
i,
int(l),
ha="center",
va="center",
color="white",
fontweight='bold',
fontsize=10)
# We want to show all ticks...
ticks = ax0.set_xticks(np.arange(len(colors.columns)))
ticks = ax0.set_yticks(np.arange(len(data['benchmark'])))
# ... and label them with the respective list entries
ticks = ax0.set_xticklabels(colors.columns)
ticks = ax0.set_yticklabels(data['benchmark'])
# Rotate the tick labels and set their alignment.
plt.setp(ax0.get_xticklabels(),
rotation=45,
ha="right",
rotation_mode="anchor")
plt.tight_layout()
plt.savefig("plot2.pdf", dpi=600)
plt.show()