Sergey Fedoseev <fedoseev.ser...@gmail.com> added the comment:
These last results are invalid :-) I thought that I was checking _PyLong_FromUnsignedChar() on top of GH-15192, but that wasn't true. So the correct results for LTO build are: $ python -m perf timeit -s "from collections import deque; consume = deque(maxlen=0).extend; b = bytes(2**20)" "consume(b)" --compare-to=../cpython-master/venv/bin/python /home/sergey/tmp/cpython-master/venv/bin/python: ..................... 6.93 ms +- 0.04 ms /home/sergey/tmp/cpython-dev/venv/bin/python: ..................... 3.96 ms +- 0.01 ms Mean +- std dev: [/home/sergey/tmp/cpython-master/venv/bin/python] 6.93 ms +- 0.04 ms -> [/home/sergey/tmp/cpython-dev/venv/bin/python] 3.96 ms +- 0.01 ms: 1.75x faster (-43%) But the most important thing is that using PyLong_FromUnsignedLong() instead of _PyLong_FromUnsignedChar() on top of GH-15192 is producing the same results: striter_next() uses small_ints[] directly. However that's not true for bytearrayiter_next(): PyLong_FromUnsignedLong() is called there. I think that's due to how code is profiled so I'm satisfied with these results more or less. I'm closing existing PR and probably will close this issue soon after GH-15192 will be merged. ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue37837> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
