Set nodeCountThreshold to 16k for >= hot compilations

[r30shah]

Set nodeCountThreshold to 16k for hot or above method compilations.

[r30shah]

@mpirvu Can I get your review on this change ?

[mpirvu]

LGTM

[r30shah]

Jenkins build all

[r30shah]

@AdamBrousseau Seems like riscV 64 build is stuck waiting for nodes which are offline (https://ci.eclipse.org/omr/job/PullRequest-linux_riscv64_cross/1400/), is there a workaround for that?

[r30shah]

@vijaysun-omr Seems like only remaining build is riscv-64 for which both machine where it can be run are offline which not sure will be resolved soon.
As it is the simple change, I have not ran internal testing builds and only have ran ODM-ILOG and Liberty for verification. Can launch ones internally if you want me to.

[vijaysun-omr]

Given the comments from Rahil, and my review of this purely heuristic change to the inliner, I feel it is very likely that the testing on RISC-V would not throw up any new issues. I am merging this change on that basis rather than hold up for infra reasons.

[JamesKingdon]

Hi @r30shah , could you give me some background on the motivation for this change? According to #7240 it looks like the larger limit would already be the default in openJ9? (for hot and above, i.e. making this a null change for openJ9)

[r30shah]

Hi @JamesKingdon , yes I believe the main intention of the change in #7240 was to lower the count for all compiles and let the consumer like OpenJ9 to set the threshold for the compiles, but we do not do that in OpenJ9 and it is only done when an environment option TR_EnableExpensiveOptsAtWarm is set which would only set higher count for warm and below. For hot and above compilations, it should have node count threshold high enough so that the resources we put in to deciding and started compiling method at hot or above opt level can get most out of it.