groovy-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jochen Theodorou <blackd...@gmx.org>
Subject Re: About the performance of the test builds
Date Sun, 29 Jan 2017 16:57:48 GMT
On 29.01.2017 15:46, Daniel Sun wrote:
> Hi all,
>
>       I noticed that test builds that use JDK 8 cost different time, they run
> tests from master and parrot branch(contains all commits from
> master)(http://ci.groovy-lang.org/project.html?projectId=Groovy), e.g.
>
> BUILD                                        TESTS COUNT            COSTED
> TIME
> JDK 8 Build#1801                     11155                         22m:35s
> JDK 8 snapshot build#995      11155                         15m:32s
> Antlr4 grammar#180               21591                         23m:48s
>
>       Who can tell me what's different among the CI instances of "JDK 8
> Build", "JDK 8 snapshot" and "Antlr4 grammar"?

hardware is the same for all. afaik the parrot branch under the "JDK 7 
build" is tested with a JDK7 build bot. "Antlr4 grammar" is tested with 
a JDK8 bot. The same bot is used for "JDK 8 build". "JDK 8 snapshot 
build" uses a checked out JDK8 to run. If "JDK 8 snapshot" is much 
faster, then maybe because of changes the JDK people did for an upcoming 
new JDK8 release

> PS:
>       If "Antlr4 grammar" CI instance has same hardware resource with "JDK 8
> Build" CI instance, we can say the new parser Parrot has better performance
> than the old one;
>       If "Antlr4 grammar" CI instance has same hardware resource with "JDK 8
> snapshot build" CI instance, we can say the new parser Parrot has almost
> same performance with the old one;

If you want to compare performance I suggest you run the build on your 
computer with different JDKs. There you have much more control over things

bye Jochen


Mime
View raw message