gh-129005: Align FileIO.readall allocation #129424
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Both now use a pre-allocated buffer of length `bufsize`, fill it using a readinto, and have matching "expand buffer" logic. On my machine this takes: `./python -m test -M8g -uall test_largefile -m test_large_read -v` from ~3.7 seconds to ~3.3 seconds
cmaloney
force-pushed
the
cmaloney/pyio_fileio_readall
branch
from
e382660 to
871979d
Compare
cmaloney
commented
Jan 29, 2025
|
Not sure how to get the Tests / Ubuntu / build and test unbroken, they seem to be auto-cancelling fast... which isn't happening on any of my other open PRs. |
|
rebased on main and opened #129458 as a replacement that hopefully will revoid the cancelled test runs. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Both
_ioand_pyionow use a pre-allocated buffer of lengthbufsize, fill it using aos.readinto/_Py_read, and have matching "expand buffer" logic.On my machine (Linux, Debug build) this takes:
./python -m test -M8g -uall test_largefile -m test_large_read -vfrom ~3.7 seconds to ~3.3 seconds_pyiostill uses 2x the memory, there are two remaining copiesbytes(result)currently copies. I'd like to either just rely on "duck typing" / bytearray is close enough to bytes, or would need to do something similar to C++ "move" semantics where the bytes could take ownership of the data buffer from the bytearray without copying... Not sure what is most Pythonic_pyio.BufferedIO._read_unlockedin the read-all case where no buffer has been allocated always doesreturn buf[:pos] + chunkwhich causes another copy. Patch for that case / "if buf is length 0, just return" coming shortly.