gh-142982: Specialize CALL_FUNCTION_EX#143391
Merged
+2,074 −1,188
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This adds a specialization for CALL_FUNCTION_EX for Python and non-Python frames.
It should be slightly faster on the interpreter, and much faster on the JIT. Previously the JIT could not trace through this operation, now it can.
JIT benchmark suggest 1.6% speedup on macOS AArch64, and 0% speedup on Linux
https://github.com/facebookexperimental/free-threading-benchmarking/blob/main/results/bm-20260102-3.15.0a3%2B-884a7a7-JIT/bm-20260102-macm4pro-arm64-Fidget%252dSpinner-call_function_ex_py-3.15.0a3%2B-884a7a7-vs-base.md
https://github.com/facebookexperimental/free-threading-benchmarking/blob/main/results/bm-20260102-3.15.0a3%2B-884a7a7-JIT/bm-20260102-vultr-x86_64-Fidget%252dSpinner-call_function_ex_py-3.15.0a3%2B-884a7a7-vs-base.md
I've checked some of the macOS speedups, and they indeed previously were blocked by CALL_FUNCTION_EX in the JIT.