The Version table provides details related to the release that this issue/RFE will be addressed.
Unresolved : Release in which this issue/RFE will be addressed. Resolved: Release in which this issue/RFE has been resolved. Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.
Currently G1 does not resize TLABs. This reduces performance for G1 since it can lead to contention in the allocation path.
Performance test show 2-3% throughput improvement when resizing TLABs up to the humongous threshold.
I have a small Java program that has two threads. One allocates a lot and one allocates very little. I did JFR recordings and looked at the object_alloc_in_new_TLAB event to see the TLAB sizes. The fast allocator gets large TLABs and the slow allocator gets small TLABs after my change.
I tried turning this in to a JTreg test. If I use JFR events to do verification I need access to the JFR parser. That means I have to put the test in among the JFR tests in the JDK. That seems like the wrong place to put a test for a Hotspot feature.
I tried using the PrintTLAB flag instead and parse the output. But that output does not contain the thread names, so I can't determine which threads that are the fast and slow allocators.