gh-115999: Enable free-threaded specialization of LOAD_CONST

[Yhg1s]

Enable free-threaded specialization of LOAD_CONST, which is an unconditional specialization without stats bookkeeping, so it's a simple opcode replace. (It could even be a relaxed store if we didn't have to worry about instrumentation.)

• Issue: Make the specializing interpreter thread-safe in --disable-gil builds #115999

[Yhg1s]

I'm not sure why LOAD_CONST was already checked off on the issue page. The test fails for free-threaded builds without the accompanying bytecode change.

[mpage]

I'm not sure why LOAD_CONST was already checked off on the issue page.

I think specializing LOAD_CONST during bytecode execution was a recent change: #128708. Previously, it happened when the bytecode was quickened during creation of the code object.

[Yhg1s]

Looks like about 1% gain from this, but given noise levels it could be a wash. Still, considering way more things are currently immortal in the free-threaded build it's probably a bigger win than in the regular build :D

https://github.com/facebookexperimental/free-threading-benchmarking/blob/main/results/bm-20250127-3.14.0a4%2B-c92089c-NOGIL/bm-20250127-vultr-x86_64-Yhg1s-more_spec-3.14.0a4%2B-c92089c-vs-base.svg

[Fidget-Spinner]

If you want a bigger win, perhaps you could allow deferred references as well. They behave the same as immortal objects if they're live on the stack.

Considering many LOAD_CONST are deferred objects, that should be a nice win in perf.

That can come in another PR though.

[Yhg1s]

Yeah, that's probably a good idea. Running some tests on the testsuite, about 75% of all LOAD_CONST invocations involved immortal objects, 25% were deferred, and the remainder is less than 1% (in the free-threaded build. Not nearly as many immortal objects in a regular build). I suspect we'll get fewer immortal objects over time, and most of them will probably be deferred instead, I guess?

Now the question is: do I add PyStackRef_FromPyObjectDeferred(), or do I change PyStackRef_FromPyObjectImmortal() to accept deferred objects as well? :)

[mpage]

LGTM!