-
Task
-
Resolution: Duplicate
-
Minor
-
None
-
None
-
None
-
NEW
-
NEW
We need to know the perf impact of JDK 9, so we can advice our customers and consultants correctly.
Create a new blog article about JDK 8 vs JDK 9 performance for optaplanner-website, similar to this one:
https://www.optaplanner.org/blog/2014/03/20/HowMuchFasterIsJava8.html
Notice how it explains the methodology, raw results and then gives an opinion.
Tip: Copy paste the adoc file of that article, see this directory: https://github.com/kiegroup/optaplanner-website/tree/master/blog
Some things to take into account:
- Add yourself in users.yml in optaplanner-website
- JDK 9 defaults to Garbage collector G1, which is a lot slower on throughput (but better on latency).
- Read and probably reference this: https://www.optaplanner.org/blog/2015/07/31/WhatIsTheFastestGarbageCollectorInJava8.html
- Try both JDK 9 with the default GC and the GC that was default in JDK 8.
- If possible, also try both with module-info.java (in all used kie, drools, optaplanner and optaplanner-examples artifacts)
- Compare it to without. Does it make a difference today? Yes/no, that's interesting info.
- Clearly note that most perf improvements from module info are coming in future versions of JDK
- You can maybe reuse the code from
PLANNER-883
- Compare it to without. Does it make a difference today? Yes/no, that's interesting info.
- duplicates
-
PLANNER-1326 Benchmark JDK 11 vs JDK 8 on GeneralOptaPlannerBenchmarkApp
- Resolved
- relates to
-
PLANNER-883 JDK 9 experiments around module-info, automatic modules, etc
- Resolved