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
180
181
182
183
184
|
import importlib.metadata
import json
import pathlib
import time
import timeit
from collections.abc import Callable, Collection, Mapping, Sequence
from dataclasses import dataclass, replace
from typing import Any, NamedTuple
import benchmarks
import matplotlib.pyplot as plt
import pandas
from common import Benchmark, Methods
ROOT_DIR = pathlib.Path(__file__).parent.parent
DATA_PATH = ROOT_DIR / "benchmark" / "data.json"
TABLE_PATH = ROOT_DIR / "examples" / "benchmark_table.md"
LIGHT_CHART_PATH = ROOT_DIR / "docs" / "benchmark_chart_light.svg"
DARK_CHART_PATH = ROOT_DIR / "docs" / "benchmark_chart_dark.svg"
CHART_TRUNCATE = 20
packages = [
path.stem
for path in pathlib.Path(benchmarks.__file__).parent.iterdir()
if not path.name.startswith("_")
]
packages = ["pydantic"]
def time_it(func: Callable, arg: Any) -> float:
timer = timeit.Timer(lambda: func(arg))
number, _ = timer.autorange()
return min(timer.repeat(number=number)) / number
def time_it_mean(func: Callable, args: Collection) -> float:
return sum(time_it(func, arg) for arg in args) / len(args)
class BenchmarkResult(NamedTuple):
first_run: float
deserialization: float
serialization: float
def run_benchmark(
methods: Methods, data: Mapping[str, Any], key: str
) -> BenchmarkResult:
print(f"\t{key}")
deserializer, serializer = methods
first_run_start = time.perf_counter_ns()
deserialized = [deserializer(elt) for elt in data[key]]
for obj in deserialized:
serializer(obj)
first_run_end = time.perf_counter_ns()
first_run = (first_run_end - first_run_start) * 1e-9
print(f"\t\tfirst run: {first_run}")
deserialization = time_it_mean(deserializer, data[key])
print(f"\t\tdeserialization: {deserialization}")
serialization = time_it_mean(serializer, deserialized)
print(f"\t\tserialization: {serialization}")
return BenchmarkResult(first_run, deserialization, serialization)
class FullBenchmarkResult(NamedTuple):
simple_deserialization: float
complex_deserialization: float
simple_serialization: float
complex_serialization: float
@dataclass(frozen=True)
class LibraryBenchmarkResult:
library: str
version: str
result: FullBenchmarkResult
def total(self) -> float:
return sum(self.result)
def relative(self, ref: "LibraryBenchmarkResult") -> "LibraryBenchmarkResult":
result = FullBenchmarkResult(*(a / b for a, b in zip(self.result, ref.result)))
return replace(self, result=result)
def run_library_benchmark(
package: str, data: Mapping[str, Any]
) -> LibraryBenchmarkResult:
print("====================")
print(package)
# import module before importing benchmark module in order to remove it
# from startup_time
importlib.import_module(package)
start_import = time.perf_counter_ns()
module = importlib.import_module(f"{benchmarks.__name__}.{package}")
end_import = time.perf_counter_ns()
startup_time = (end_import - start_import) * 1e-9
simple_methods, complex_methods, library = next(
val for val in module.__dict__.values() if isinstance(val, Benchmark)
)
library = library or package
simple_results, complex_results = [
run_benchmark(methods, data, key)
for methods, key in [(simple_methods, "simple"), (complex_methods, "complex")]
]
print(
f"startup time: {startup_time + simple_results.first_run + complex_results.first_run}"
)
return LibraryBenchmarkResult(
library,
importlib.metadata.version(library),
FullBenchmarkResult(
simple_results.deserialization,
complex_results.deserialization,
simple_results.serialization,
complex_results.serialization,
),
)
def export_table(results: Sequence[LibraryBenchmarkResult]):
with open(TABLE_PATH, "w") as table:
table.write("|library|version|deserialization|serialization|\n")
table.write("|-|-|-:|-:|\n")
for res in results:
if all(r == 1.0 for r in res.result):
deserialization, serialization = "/", "/"
else:
deserialization, serialization = [
f"x{round(sum(res.result[index:index + 2]) / 2, 1)}"
f" ({round(res.result[index], 1)}/{round(res.result[index + 1], 1)})"
for index in (0, 2)
]
table.write(
f"|{res.library}|{res.version}|{deserialization}|{serialization}|\n"
)
def export_chart(
results: Sequence[LibraryBenchmarkResult], path: pathlib.Path, style: str
):
plt.style.use(style)
columns = [
f"{op} ({bench})"
for op in ("deserialization", "serialization")
for bench in ("simple", "complex")
]
# I've used pandas because I was not able to do what I wanted with matplotlib alone
df = pandas.DataFrame(
[
[res.library] + [min(r, CHART_TRUNCATE) for r in res.result]
for res in results
],
columns=["library"] + columns,
)
ax = df.plot.bar(x="library", title="Benchmark (lower is better)", rot=45)
ax.legend(framealpha=0, loc="upper left")
plt.xlabel("")
plt.tight_layout()
for container in ax.containers:
ax.bar_label(
container,
labels=["" if v < CHART_TRUNCATE else "ยทยท" for v in container.datavalues],
padding=2,
rotation=90,
)
plt.savefig(str(path), transparent=True)
def main():
with open(DATA_PATH) as json_file:
data = json.load(json_file)
results = sorted(
(run_library_benchmark(p, data) for p in packages),
key=LibraryBenchmarkResult.total,
)
relative_results = [res.relative(results[0]) for res in results]
export_table(relative_results)
export_chart(relative_results, LIGHT_CHART_PATH, "default")
export_chart(relative_results, DARK_CHART_PATH, "dark_background")
if __name__ == "__main__":
main()
|
