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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
|
from __future__ import print_function
from __future__ import division
import argparse
from enum import Enum
import xml.etree.ElementTree as ET
from collections import OrderedDict
import matplotlib
import matplotlib.pyplot as plt
class BenchmarkResult:
"""BenchmarkResult class to hold perf numbers"""
def __init__(self, name, mean_value, mean_lower_bound, mean_upper_bound, \
std_val, std_lower_bound, std_upper_bound):
self.name = name
self.mean_value = mean_value
self.mean_lower_bound = mean_lower_bound
self.mean_upper_bound = mean_upper_bound
self.std_val = std_val
self.std_lower_bound = std_lower_bound
self.std_upper_bound = std_upper_bound
def __str__(self):
return "BenchmarkResults %s\nmean value=%f, lowerBound=%f, upperBound=%f\nstd:%f, lowerBound=%f, upperBound=%f" \
% (self.name, self.mean_value, self.mean_lower_bound, self.mean_upper_bound, \
self.std_val, self.std_lower_bound, self.std_upper_bound)
class TestCase:
"""TestCase class to hold the list of benchmark results"""
def __init__(self, name):
self.name = name
self.results = OrderedDict()
def __str__(self):
return "TestCase %s\nBenchmark result:\n%s" % (self.name, self.results)
class SectionCase:
"""SectionCase class to hold the section list of benchmark results"""
def __init__(self, name):
self.name = name
self.results = OrderedDict()
def __str__(self):
return "SectionCase %s\nBenchmark result:\n%s" % (self.name, self.results)
class Metric(Enum):
mean = 'mean'
lowMean = 'low_mean'
highMean = 'high_mean'
class Unit(Enum):
nano = 'nano'
micro = 'micro'
milli = 'milli'
sec = 'sec'
def displayUnit(unit):
if unit == Unit.sec:
return "s"
elif unit == Unit.milli:
return "ms"
elif unit == Unit.micro:
return "us"
else:
return "ns"
def nanoToMicro(nano):
return nano / 1000
def nanoToMilli(nano):
return nano / (1000*1000)
# return nano
def nanoToSec(nano):
return nano / (1000*1000*1000)
def convertUnit(nano, unit):
if unit == Unit.sec:
return nanoToSec(nano)
elif unit == Unit.milli:
return nanoToMilli(nano)
elif unit == Unit.micro:
return nanoToMicro(nano)
else:
return nano
parser = argparse.ArgumentParser()
parser.add_argument("--xml-file", help="Path to XML perf log.", required=True)
parser.add_argument("--label", help="Label for before column.", default='Before')
parser.add_argument("--metric", help="Benchmark metric (mean, low_mean, high_mean).", type=Metric, choices=list(Metric), default=Metric.mean)
parser.add_argument("--unit", help="Benchmark unit (nano, micro, milli, sec).", type=Unit, choices=list(Unit), default=Unit.milli)
args = parser.parse_args()
xml_file = args.xml_file
metric = args.metric
time_unit = args.unit
print("Matplotlib: {}".format(matplotlib.__version__))
print('Path to XML log file for perf comparison:', xml_file)
print('Perf metric:', metric)
print('Time unit:', time_unit)
print()
tree_perf_data = ET.parse(xml_file)
root_perf_data = tree_perf_data.getroot()
def loadResults(root):
results = OrderedDict()
for test_case in root.iter('TestCase'):
test_name = test_case.attrib['name']
current_test = TestCase(test_name)
for section_case in test_case.iter('Section'):
case_name = section_case.attrib['name']
current_section = SectionCase(case_name)
for bench_res in section_case.iter('BenchmarkResults'):
bench_name = bench_res.attrib['name']
mean_node = bench_res.find('mean')
mean = float(mean_node.attrib['value'])
mean_lower = float(mean_node.attrib['lowerBound'])
mean_upper = float(mean_node.attrib['upperBound'])
std_node = bench_res.find('standardDeviation')
std = float(std_node.attrib['value'])
std_lower = float(std_node.attrib['lowerBound'])
std_upper = float(std_node.attrib['upperBound'])
current_section.results[bench_name] = BenchmarkResult(bench_name, mean, mean_lower, mean_upper, std, std_lower, std_upper)
current_test.results[case_name] = current_section
results[test_name] = current_test
return results
results_perf_data = loadResults(root_perf_data)
for r_name, r_test in results_perf_data.items():
fig, axs = plt.subplots(ncols=len(r_test.results.items()))
idx1 = 0
for r_section_name, r_section in r_test.results.items():
x_list = []
y_list = []
std_list = []
idx2 = 0
backend_list = []
for r_result_name, r_result in r_section.results.items():
x_list.append(idx2)
if metric == Metric.lowMean:
y_list.append(convertUnit(r_result.mean_lower_bound, time_unit))
elif metric == Metric.highMean:
y_list.append(convertUnit(r_result.mean_higher_bound, time_unit))
else:
y_list.append(convertUnit(r_result.mean_value, time_unit))
std_list.append(convertUnit(r_result.std_val, time_unit))
backend_list.append(r_result_name.replace(" backend", ""))
idx2 += 1
axs[idx1].bar(x_list, y_list, yerr=std_list, align='center', capsize=10, tick_label=backend_list)
axs[idx1].grid(True)
axs[idx1].xaxis.set_tick_params(labelsize=14)
axs[idx1].set_xlabel(r_section_name, fontsize=16)
axs[0].set_ylabel("Computation time (ms)", fontsize=16)
plt.setp(axs[idx1].get_xticklabels(), rotation=45)
idx1 += 1
plt.suptitle(r_name, fontsize=24)
plt.show()
|
