RFR: 8377597: [Leyden] Improve peak performance when AOT code is used [v2]

[Aleksey Shipilev]

On Thu, 12 Feb 2026 17:24:28 GMT, Vladimir Kozlov <kvn at openjdk.org> wrote:

> Counters increment is not synchronized and not atomic. Which is similar to other profiling counters we have.

Yes, and where we _do_ have a counter contention problem. In mainline, it is constrained to T2/T3 code, which is more or less transient. Alas, doing the same in A4 would expose this problem to "optimal" code. Andrew is working on [JDK-8372701](https://bugs.openjdk.org/browse/JDK-8372701), which might be of benefit there. But none of that code deals with counter updates on C2 side, so there are some obstacles to overcome.

-------------

PR Comment: https://git.openjdk.org/leyden/pull/110#issuecomment-3892306814