Skip to main content



      Home
Home » Eclipse Projects » Eclipse Platform » org.eclipse.test.performance usage
org.eclipse.test.performance usage [message #284175] Tue, 19 April 2005 10:21 Go to next message
Eclipse UserFriend
Hello,

I've been trying to use the eclipse performance framework to get a continual
pulse on some of the features in CDT.

I am subclassing PerformanceTestCase and I am seeing that I am getting
negative values from time to time wrt memory. I'm wondering if there is
something I have missed from the HowTo doc that I'm supposed to do to reset
the memory counters properly.
When I run one test stand-alone, the counters all work out positive, but if
I run a few tests in a suite, I start getting negative values.

Thanks
JohnC

Scenario
'org.eclipse.cdt.core.tests.performance.parser.ASTServiceCPP ParserTest#testT
rilogyLocationPerformance()' (average over 20 samples):
Used Java Heap: 609.09K
Working Set: 1.6K
Committed: -819
Working Set Peak: 0
Elapsed Process: 109 ms
Kernel time: 0 ms
Page Faults: 0
CPU Time: 109 ms
GDI Objects: 0
Re: org.eclipse.test.performance usage [message #284198 is a reply to message #284175] Tue, 19 April 2005 13:52 Go to previous messageGo to next message
Eclipse UserFriend
Originally posted by: john.eclipsefaq.org

If you look through SDK performance test results, you'll see the same
thing. This is showing the memory usage delta between the snapshot at
the beginning and the snapshot at the end. If the garbage collector
runs during your test, for example, you can end up with negative
numbers. I agree this is misleading; essentially this information is
not useful for most tests. For tests where you really want to measure
the memory growth, I suggest experimenting with running GCs before
starting your timer to see if this stabilizes your results.
--

johnc wrote:
> Hello,
>
> I've been trying to use the eclipse performance framework to get a continual
> pulse on some of the features in CDT.
>
> I am subclassing PerformanceTestCase and I am seeing that I am getting
> negative values from time to time wrt memory. I'm wondering if there is
> something I have missed from the HowTo doc that I'm supposed to do to reset
> the memory counters properly.
> When I run one test stand-alone, the counters all work out positive, but if
> I run a few tests in a suite, I start getting negative values.
>
> Thanks
> JohnC
>
> Scenario
> 'org.eclipse.cdt.core.tests.performance.parser.ASTServiceCPP ParserTest#testT
> rilogyLocationPerformance()' (average over 20 samples):
> Used Java Heap: 609.09K
> Working Set: 1.6K
> Committed: -819
> Working Set Peak: 0
> Elapsed Process: 109 ms
> Kernel time: 0 ms
> Page Faults: 0
> CPU Time: 109 ms
> GDI Objects: 0
>
>
Re: org.eclipse.test.performance usage [message #284291 is a reply to message #284198] Wed, 20 April 2005 09:55 Go to previous message
Eclipse UserFriend
Thanks

"John Arthorne" <john@eclipsefaq.org> wrote in message
news:d43h0h$ohe$1@news.eclipse.org...
> If you look through SDK performance test results, you'll see the same
> thing. This is showing the memory usage delta between the snapshot at
> the beginning and the snapshot at the end. If the garbage collector
> runs during your test, for example, you can end up with negative
> numbers. I agree this is misleading; essentially this information is
> not useful for most tests. For tests where you really want to measure
> the memory growth, I suggest experimenting with running GCs before
> starting your timer to see if this stabilizes your results.
> --
>
> johnc wrote:
> > Hello,
> >
> > I've been trying to use the eclipse performance framework to get a
continual
> > pulse on some of the features in CDT.
> >
> > I am subclassing PerformanceTestCase and I am seeing that I am getting
> > negative values from time to time wrt memory. I'm wondering if there is
> > something I have missed from the HowTo doc that I'm supposed to do to
reset
> > the memory counters properly.
> > When I run one test stand-alone, the counters all work out positive, but
if
> > I run a few tests in a suite, I start getting negative values.
> >
> > Thanks
> > JohnC
> >
> > Scenario
> >
'org.eclipse.cdt.core.tests.performance.parser.ASTServiceCPP ParserTest#testT
> > rilogyLocationPerformance()' (average over 20 samples):
> > Used Java Heap: 609.09K
> > Working Set: 1.6K
> > Committed: -819
> > Working Set Peak: 0
> > Elapsed Process: 109 ms
> > Kernel time: 0 ms
> > Page Faults: 0
> > CPU Time: 109 ms
> > GDI Objects: 0
> >
> >
Previous Topic:How to complie plugins/RCP outside eclipse
Next Topic:About Editor,why don't provide drag and drop ability for text blocks?
Goto Forum:
  


Current Time: Fri Sep 26 22:32:44 EDT 2025

Powered by FUDForum. Page generated in 0.05248 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top