blob: 2a62f56080ae460812b2a241cdef2d6440a4d774 (
plain)
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
|
# $Id$
Performance tests for the real-time event channel.
This directory contains several tests to measure the performance of
the real-time event channel. Remember to compile ACE and TAO with
optimizations enabled when running this tests.
Here are some canonical test configurations:
# Measure throughput and latency, in a configuration suitable for
# single threaded systems. This is probably the best case scenario,
# one supplier, one consumer and no locking.
$ Throughput -burstsize 100000 -burstcount 1
# Same test as above, but with a configuration that is MT-safe
$ Throughput -ORBsvcconf ec.st.conf -burstsize 100000 -burstcount 1
# Increase the number of consumers, but only the first one receives
# events.
$ Throughput -ORBsvcconf ec.st.conf -burstsize 100000 -burstcount 1 \
-consumers 4
# Now increase the number of suppliers, only one consumer receives
# events from each supplier.
$ Throughput -ORBsvcconf ec.st.conf -burstsize 100000 -burstcount 1 \
-consumers 4 -suppliers 4
# Now a test where all consumers receive events from all suppliers
$ Throughput -ORBsvcconf ec.st.conf -burstsize 100000 -burstcount 1 \
-consumers 4 -suppliers 4 -consumers_tshift 0 -suppliers_tshift 0
# Measure the time required to connect the 100 consumers and 100
# suppliers, inserting one supplier first and then one supplier.
# Try -connection_order "consumers" or "suppliers"
$ Connect -ORBsvcconf ec.st.conf -consumers 100 -suppliers 100 \
-connection_order interleaved
NOTES
Don't worry about the "incomplete data" warning, it is a
deffect in the test.
|