René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Add return const instruction","articleBody":"From the pystats doc ([pystats-2023-02-05-python-5a2b984.md](https://github.com/faster-cpython/ideas/blob/main/stats/pystats-2023-02-05-python-5a2b984.md)), I find that `LOAD_CONST + RETURN_VALUE` is a very high frequency (Because the default return of the function is None).\r\n\r\n\r\nSuccessors for LOAD_CONST \r\n\r\nSuccessors | Count | Percentage\r\n-- | -- | --\r\nRETURN_VALUE | 969,173,651 | 21.8%\r\nBINARY_OP_ADD_INT | 418,647,997 | 9.4%\r\nLOAD_CONST | 403,185,774 | 9.1%\r\nCOMPARE_AND_BRANCH_INT | 314,633,792 | 7.1%\r\nSTORE_FAST | 295,563,626 | 6.6%\r\n\r\nAnd predecessors for RETURN_VALUE\r\n\r\nPredecessors | Count | Percentage\r\n-- | -- | --\r\nLOAD_CONST | 969,173,651 | 29.9%\r\nLOAD_FAST | 505,933,343 | 15.6%\r\nRETURN_VALUE | 382,698,373 | 11.8%\r\nBUILD_TUPLE | 328,532,240 | 10.1%\r\nCOMPARE_OP | 107,210,803 | 3.3%\r\n\r\nThis means that if we add a `RETURN_CONST`, we can reduce the `RETURN_VALUE` instruction by 30% and the `LOAD_CONST` instruction by 20%.\r\n\r\n\r\n```\r\n./bin/python3 -m pyperf timeit -w 3 --compare-to ../python-3.12/bin/python3 -s \"\r\ndef test():\r\n    return 10000\r\n\" \"test()\"\r\n\r\n/python-3.12/bin/python3: ..................... 27.0 ns +- 0.3 ns\r\n/cpython/bin/python3: ..................... 25.0 ns +- 0.5 ns\r\nMean +- std dev: [/python-3.12/bin/python3] 27.0 ns +- 0.3 ns -\u003e [/cpython/bin/python3] 25.0 ns +- 0.5 ns: 1.08x faster\r\n\r\n./bin/python3 -m pyperf timeit -w 3 --compare-to ../python-3.12/bin/python3 -s \"\r\ndef test():\r\n    return None\r\n\" \"test()\"\r\n\r\n/python-3.12/bin/python3: ..................... 27.2 ns +- 1.3 ns\r\n/cpython/bin/python3: ..................... 25.1 ns +- 0.6 ns\r\nMean +- std dev: [/python-3.12/bin/python3] 27.2 ns +- 1.3 ns -\u003e [/cpython/bin/python3] 25.1 ns +- 0.6 ns: 1.08x faster\r\n```\r\nFrom the microbenchmark that there is indeed a ~10% improvement (considering the interference of function calls, I think 10% should be there), which is not very high, but it should be an optimization without adverse effects.\r\n\n\n\u003c!-- gh-linked-prs --\u003e\n### Linked PRs\n* gh-101633\n\u003c!-- /gh-linked-prs --\u003e\n","author":{"url":"https://github.com/penguin-wwy","@type":"Person","name":"penguin-wwy"},"datePublished":"2023-02-07T06:53:47.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":2},"url":"https://github.com/101632/cpython/issues/101632"}