Bug 311546 - [perfs] Regression for StatePerformanceTest#testResolution100()
Summary: [perfs] Regression for StatePerformanceTest#testResolution100()
Status: RESOLVED FIXED
Alias: None
Product: Equinox
Classification: Eclipse Project
Component: Framework (show other bugs)
Version: 3.6   Edit
Hardware: PC Windows XP
: P3 normal (vote)
Target Milestone: 3.6 RC1   Edit
Assignee: equinox.framework-inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords: test
Depends on:
Blocks: 286955
  Show dependency tree
 
Reported: 2010-05-04 09:36 EDT by Frederic Fusier CLA
Modified: 2010-05-21 07:01 EDT (History)
2 users (show)

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Frederic Fusier CLA 2010-05-04 09:36:36 EDT 
Verifying 3.6M7 perf results, it appears that I missed to report a regression on StatePerformanceTest#testResolution100() test which occurs since build N20100415-2000 (between -14% and -20%).

Note that I also can see a similar regression (ie. which appeared exactly on the same build) on following tests:
 - StateUsesPerformanceTest#testUsesResolution00100()
 - StateUsesPerformanceTest#testUsesResolution00500()

This regression is only noticeable on the Win XP Sun 1.5.0_10 test machine.

Sorry, I'm not sure it's the right component, please reassign if necessary, thanks
Comment 1 Thomas Watson CLA 2010-05-04 10:03:05 EDT 
This looks to be caused by the fix in bug269185.  We are trading off reduction of memory usage for a small jump in execution time.
Comment 2 Frederic Fusier CLA 2010-05-04 10:27:27 EDT 
I also can see a similar regression (ie. which appeared exactly on
the same build) on following test:
 - StatePerformanceTest#testResolution500()

My tool didn't notice me for it as it's just at the -10% threshold, sometimes
over and sometimes below, hence there are never 3 consecutive builds having
this failure...
Comment 3 Frederic Fusier CLA 2010-05-04 10:28:32 EDT 
(In reply to comment #1)
> This looks to be caused by the fix in bug269185.  We are trading off reduction
> of memory usage for a small jump in execution time.

Hence, the best solution in this case is to add a comment in these tests.
Comment 4 Thomas Watson CLA 2010-05-04 11:12:39 EDT 
I added a regression comment to the four tests you have identified.