8381834: Improve TLAB sizing heuristics

[albertnetymk]

The patch introduces global allocation fraction tracking (AdaptiveWeightedAverage) alongside per-thread allocation fraction. It uses observed global requested-size fraction to estimate initial per-thread TLAB size. Additionally, there's a temporary size boost (up to 16X) when refill pressure is high so threads with real allocation demand do not regress.

I ran MTLinkedListGC.java before and after the change. The largest effect is in debug builds: both Parallel and G1 end up with a much smaller heap, and many fewer GCs. Release builds also improve, although the gap is smaller there.

[linux-x64-debug]
ParallelGC
baseline: [168.731s][info][gc] GC(636) Pause Young (Allocation Failure) 877M->150M(880M) 86.101ms
patched : [3.917s][info][gc] GC(22) Pause Young (Allocation Failure) 23M->9M(27M) 3.579ms

G1GC
baseline: [40.078s][info][gc] GC(80) Pause Young (Normal) (G1 Evacuation Pause) 1471M->9M(7960M) 199.514ms
patched : [3.298s][info][gc] GC(10) Pause Young (Normal) (G1 Evacuation Pause) 126M->23M(180M) 16.209ms

Performance-neutral for DaCapo, SPECjbb2015, and SPECjvm2008. Also checked in gc+tlab=debug logs that the number of tlab refills are mostly on par.

Test: tier1-5

---

Progress

• Change must be properly reviewed (1 review required, with at least 1 Reviewer)
• Change must not contain extraneous whitespace
• Commit message must refer to an issue

Issue

• JDK-8381834: Improve TLAB sizing heuristics (Enhancement - P4)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/30622/head:pull/30622
$ git checkout pull/30622

Update a local copy of the PR:
$ git checkout pull/30622
$ git pull https://git.openjdk.org/jdk.git pull/30622/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 30622

View PR using the GUI difftool:
$ git pr show -t 30622

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/30622.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back ayang! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

[openjdk]

❗ This change is not yet ready to be integrated.
See the Progress checklist in the description for automated requirements.

[openjdk]

@albertnetymk The following label will be automatically applied to this pull request:

• hotspot-gc

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

[mlbridge]

Webrevs

• 00: Full (c9f80695)

[walulyai]

For the log snippet showing the improvement, is this with ~1000 threads as described in the CR?

[albertnetymk]

For the log snippet showing the improvement, is this with ~1000 threads as described in the CR?

Yes, from running MTLinkedListGC.java with -Xlog:gc. That test creates ~1000 threads.