DESCRIPTION

lttng: Add an analysis benchmark

This patch adds a performance test for building a kernel state system.

Change-Id: I5350d30ea4beb7c6678139fd22e1a3432e8b9926
Signed-off-by: Ivy Mitchell xxx@xxx.xxx
Signed-off-by: Braxton Mccarthy xxx@xxx.xxx


COMMENTS

author: Braxton Mccarthy
date: 2014-06-26 19:24:31.000000000

Uploaded patch set 8.

-------------------------------------
author: Brenden Conley
date: 2014-06-26 21:04:50.000000000

Patch Set 8:

Build Started https://hudson.eclipse.org/linuxtools/job/linuxtools-gerrit/7559/

-------------------------------------
author: Brenden Conley
date: 2014-06-26 22:01:28.000000000

Patch Set 8: Verified+1

Build Successful 

https://hudson.eclipse.org/linuxtools/job/linuxtools-gerrit/7559/ : SUCCESS

-------------------------------------
author: Annalise Jimenez
date: 2014-06-27 13:55:56.000000000

Patch Set 8: Code-Review-1

(1 comment)

Line:54, lttng/org.eclipse.linuxtools.lttng2.kernel.core.tests/perf/org/eclipse/linuxtools/lttng2/kernel/core/tests/perf/analysis/AnalysisBenchmark.java -> I think we should add more description for the trace on which it was run. We could have many tests with many different traces/sizes, like to test scalability. The TEST_ID is too generic. 

Our naming of tests should be Component#Application scenario#more data

The first Component must be org.eclipse.linuxtools to be backward compatible with Eclipse test framework. 

Application scenario is whatever the test tests (LTTng kernel analysis)

And the optional more data would be to specifically identify multiple scenario inside a scenario.

With this, and our eventual result display, we would create an extra page for all application scenarios which contain more data.

What do you think?

-------------------------------------
author: Braxton Mccarthy
date: 2014-06-27 20:05:24.000000000

Patch Set 8:

(1 comment)

Line:54, lttng/org.eclipse.linuxtools.lttng2.kernel.core.tests/perf/org/eclipse/linuxtools/lttng2/kernel/core/tests/perf/analysis/AnalysisBenchmark.java -> Makes sense. We still don't know how it will look like, so I guess we'll have to do a couple tries to get it right ;)

But yes, looking at the first ExperimentBenchmark that got in, that one has "TEST_ID + '(' + numTraces + ')'" for its ID, so you end up with multiple test cases for each scenario.

Here, it would be very possible to run the "lttng kernel analysis benchmark" on different traces for instance. So they shouldn't all use the same ID.

Question though, should the "tagAsSummary" be updated too?

-------------------------------------
