summaryrefslogtreecommitdiff
path: root/trunk/TAO/performance-tests/Latency/Single_Threaded/README
diff options
context:
space:
mode:
Diffstat (limited to 'trunk/TAO/performance-tests/Latency/Single_Threaded/README')
-rw-r--r--trunk/TAO/performance-tests/Latency/Single_Threaded/README26
1 files changed, 26 insertions, 0 deletions
diff --git a/trunk/TAO/performance-tests/Latency/Single_Threaded/README b/trunk/TAO/performance-tests/Latency/Single_Threaded/README
new file mode 100644
index 00000000000..e1f411e2632
--- /dev/null
+++ b/trunk/TAO/performance-tests/Latency/Single_Threaded/README
@@ -0,0 +1,26 @@
+/**
+
+@page Single Threaded Performance Test README File
+
+ This test tries to estimate the minimum latency for a twoway
+request. The test uses a single threaded client and server, and
+configures the ORB to eliminate common sources of overhead, such as
+locking.
+
+ Please do not extend this test to deal with other data types,
+configurations, etc. If you need to just create a new test. In the
+past we had multi-purpose tests like that, they were hard to maintain,
+and the results were hard to interpret. Furthermore, the test is
+purposely simple so its output can be parsed via nightly scripts and
+plotted in a graph. More importantly, the overhead of marshaling is
+better measures using an isolated test, that does not involve the
+network.
+
+ To run the test use the run_test.pl script:
+
+$ ./run_test.pl
+
+ the script returns 0 if the test was successful, and prints
+out the performance numbers.
+
+*/