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
185
186
187
188
|
"""
Measure performance of pubsub, so that impact of proposed performance enhancing
algorithms can be proven. Measure with
python -m timeit -n1 "import perf; perf.runTest()"
:copyright: Copyright since 2006 by Oliver Schoenborn, all rights reserved.
:license: BSD, see LICENSE.txt for details.
"""
from pathlib import Path
from time import perf_counter
from typing import Tuple
from pubsub import pub
from pubsub.core import Listener
topicMgr = pub.getDefaultTopicMgr()
def loop_subscribe(topic_names, listeners):
topicMgr = pub.getDefaultTopicMgr()
topicMgr.clearTree()
num_samples = 1000
start = perf_counter()
for count in range(num_samples):
for listener in listeners:
for topic_name in topic_names:
pub.subscribe(listener, topic_name)
tot_time = (perf_counter() - start)
return round(tot_time, 2)
def perf_subscribe():
"""
Subscription involves inspecting listener to (upon first subscription to a topic) create the
topic and its MDS or (upon subsequent subscriptions to same topic) verify its signature is
compatible with topic MDS, then adding listener to topic listeners, and adding topic to
topics mapping (of name to topic object). This suggests that complexity of signature and
complexity of topic name are the two axes of performance for subscriptions, as the
creation of a topic is not likely to be nearly as common as subscriptions to existing topics.
Results indicate that deeper topics and listeners have only marginal (ie insignificant)
overhead over shallow topics and listeners:
- root topic and no-args listeners: 6.77
- deep topics and no-args listeners: 6.83
- root topics and many-args listeners: 6.8
- deep topics and many-args listeners: 6.83
"""
print("-"*40)
print("Performance measurement for subscribing:")
num_topics = 100
num_listeners = 10
# root topic and no-args listeners:
topic_names = ['some_topic_' + str(topic_index) for topic_index in range(num_topics)]
listeners = [lambda: n for n in range(num_listeners)]
print('root topic and no-args listeners:', loop_subscribe(topic_names, listeners))
# deep topics and no-args listeners:
topic_names = ['rt.st.sst.ssst.leaf_topic_' + str(topic_index) for topic_index in range(num_topics)]
print('deep topics and no-args listeners:', loop_subscribe(topic_names, listeners))
# root topics and many-args listeners:
topic_names = ['some_topic_' + str(topic_index) for topic_index in range(num_topics)]
listeners = [lambda x, y, z, a, b, c: n for n in range(num_listeners)]
print('root topics and many-args listeners:', loop_subscribe(topic_names, listeners))
# deep topics and many-args listeners:
topic_names = ['rt.st.sst.ssst.leaf_topic_' + str(topic_index) for topic_index in range(num_topics)]
print('deep topics and many-args listeners:', loop_subscribe(topic_names, listeners))
def loop_send(subscriptions: Tuple[Listener, str], messages):
topicMgr = pub.getDefaultTopicMgr()
topicMgr.clearTree()
for listener, topic_name in subscriptions:
pub.subscribe(listener, topic_name)
num_samples = 1000
start = perf_counter()
for count in range(num_samples):
for topic_name, kwargs in messages:
pub.sendMessage(topic_name, **kwargs)
tot_time = (perf_counter() - start)
return round(tot_time, 2)
def perf_send():
"""
Sending message involves calling each listener with data, and going up the tree to root actor
each level has fewer data so data must be filtered out.
----------------------------------------
Performance measurement for sending:
listeners ['obs1', 'obs2', 'obs3', 'obs4', 'obs5', 'obs6', 'obs7', 'obs8']
topic names ['t1', 't1.t2', ..., 't1.t2.t3.t4.t5.t6.t7.t8']
with depth 8: with depth 8: % less X faster
8 data 9.31 8 data 4.76 49 2.0
4 data 8.48 4 data 4.48 47 1.9
2 data 8.23 2 data 4.5 45 1.8
1 data 8.26 1 data 4.31 48 1.9
no data 8.05 no data 4.27 47 1.8
with depth 4: with depth 4:
4 data 4.4 4 data 2.63 40 1.7
2 data 3.96 2 data 2.56 35 1.5
1 data 3.99 1 data 2.51 37 1.6
no data 4.1 no data 2.43 40 1.7
with depth 2: with depth 2:
2 data 2.28 2 data 1.61 29 1.4
1 data 1.84 1 data 1.57 15 1.2
no data 2.18 no data 1.52 30 1.4
with depth 1: with depth 1:
1 data 1.2 1 data 1.1 8 1.1
no data 1.24 no data 1.05 15 1.2
"""
print("-"*40)
print("Performance measurement for sending:")
# root topic and no-args listeners:
def obs1(arg1=None): pass
def obs2(arg1=None, arg2=None): pass
def obs3(arg1=None, arg2=None, arg3=None): pass
def obs4(arg1=None, arg2=None, arg3=None, arg4=None): pass
def obs5(arg1=None, arg2=None, arg3=None, arg4=None, arg5=None): pass
def obs6(arg1=None, arg2=None, arg3=None, arg4=None, arg5=None, arg6=None): pass
def obs7(arg1=None, arg2=None, arg3=None, arg4=None, arg5=None, arg6=None, arg7=None): pass
def obs8(arg1=None, arg2=None, arg3=None, arg4=None, arg5=None, arg6=None, arg7=None, arg8=None): pass
local_objs = locals().copy()
listeners = [local_objs['obs' + str(n)] for n in range(1, 9)]
print('listeners', [cb.__name__ for cb in listeners])
topic_names = ['t1']
for index in range(2, 9):
topic_names.append(topic_names[-1] + ".t" + str(index))
print('topic names', topic_names)
num_messages = 100
def sub_test(topic_names):
subscriptions = [(obs, name) for obs, name in zip(listeners, topic_names)]
num_topics = len(topic_names)
num_subs = len(subscriptions)
print('with depth {}:'.format(num_topics))
if len(topic_names) >= 8:
msg_data = dict(arg1=1, arg2=2, arg3=3, arg4=4, arg5=5, arg6=6, arg7=7, arg8=8)
messages = [(topic_names[-1], msg_data)] * num_messages
print(' {} data'.format(len(msg_data)), loop_send(subscriptions, messages))
if len(topic_names) >= 4:
msg_data = dict(arg1=1, arg2=2, arg3=3, arg4=4)
messages = [(topic_names[-1], msg_data)] * num_messages
print(' {} data'.format(len(msg_data)), loop_send(subscriptions, messages))
if len(topic_names) >= 2:
msg_data = dict(arg1=1, arg2=2)
messages = [(topic_names[-1], msg_data)] * num_messages
print(' {} data'.format(len(msg_data)), loop_send(subscriptions, messages))
if len(topic_names) >= 1:
msg_data = dict(arg1=1)
messages = [(topic_names[-1], msg_data)] * num_messages
print(' {} data'.format(len(msg_data)), loop_send(subscriptions, messages))
messages = [(topic_names[-1], {})] * num_messages
print(' {} data'.format('no'), loop_send(subscriptions, messages))
sub_test(topic_names)
sub_test(topic_names[:4])
sub_test(topic_names[:2])
sub_test(topic_names[:1])
if __name__ == '__main__':
perf_subscribe()
perf_send()
|
