-
Bug
-
Resolution: Fixed
-
P4
-
None
-
b20
In G1 code, the `G1ConfidencePercent` option seems to be interpreted inconsistently:
- for predictions it looks like smaller values make the results closer to the average of the samples, indicating more trust/confidence in them
- for calculating the expected safety margin for the size of the copied bytes in young gen. There, smaller values make the margin larger, i.e. less trust in other predictions.
Unify that.
- for predictions it looks like smaller values make the results closer to the average of the samples, indicating more trust/confidence in them
- for calculating the expected safety margin for the size of the copied bytes in young gen. There, smaller values make the margin larger, i.e. less trust in other predictions.
Unify that.
- links to
-
Commit(master) openjdk/jdk/3b8a2f8c
-
Review(master) openjdk/jdk/21448