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

[Vladimir Kozlov]

On Thu, 12 Feb 2026 17:13:58 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> ...which is a bit weird, because I do see this work well as recompilation policy solution: the bulk of the code is replaced by T4 by the end of the run:

@shipilev,  thank you for this data. We also observed something similar: almost all AOT code is replaced but still small regression 5-10% is present at peak.

I/we suspect it could be for several reasons - mostly because new C2 code is not optimal as without AOT. It could be due to different inlining decisions or other things.  We will investigate this after this PR.

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

PR Comment: https://git.openjdk.org/leyden/pull/110#issuecomment-3892321766