Details
Description
I have designed a "STW" garbage collection for CMP. Its total pause time is 24.9% to 62.1% shorter than that of the garbage collection in Harmony M2 on Intel dual-core platform. Its performance is improved 20% in average. The detail experimental result is in the attachment. (My development environment is 5.0M2).
In the attachment there are the souce code, compile configure file and test report.
The following are the two parameters In my garbage collection.
-XX:gc.corenum=core number default value is 2 i.e -XX:gc.corenum=2
-XX:gc.blocksize=block size default value is 64 i.e -XX:gc.blocksize=64
I am very glad to get some comments from everyone.
Thank you.