Description

assignee = None
closed_at = None
created_at = <Date 2021-06-06.22:05:07.766>
labels = ['3.11', 'library', 'OS-windows', 'performance']
title = 'time.monotonic() should use a different clock source on Windows'
updated_at = <Date 2021-06-14.21:10:46.833>
user = 'https://github.com/lunixbochs'

activity = <Date 2021-06-14.21:10:46.833>
actor = 'eryksun'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)', 'Windows']
creation = <Date 2021-06-06.22:05:07.766>
creator = 'lunixbochs2'
dependencies = []
files = []
hgrepos = []
issue_num = 44328
keywords = []
message_count = 12.0
messages = ['395221', '395238', '395490', '395493', '395681', '395683', '395719', '395769', '395771', '395782', '395784', '395849']
nosy_count = 9.0
nosy_names = ['paul.moore', 'belopolsky', 'vstinner', 'tim.golden', 'zach.ware', 'eryksun', 'steve.dower', 'p-ganssle', 'lunixbochs2']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue44328'
versions = ['Python 3.11']

Activity

1. lunixbochs commented on Jun 6, 2021

   

   lunixbochsmannequin

   on Jun 6, 2021

   MannequinAuthor

   More actions

   Related to https://bugs.python.org/issue41299#msg395220

   Presumably time.monotonic() on Windows historically used GetTickCount64() because QueryPerformanceCounter() could fail. However, that hasn't been the case since Windows XP: https://docs.microsoft.com/en-us/windows/win32/api/profileapi/nf-profileapi-queryperformancecounter

   On systems that run Windows XP or later, the function will always succeed and will thus never return zero

   I've run into issues with this when porting python-based applications to Windows. On other platforms, time.monotonic() was a decent precision so I used it. When I ported to Windows, I had to replace all of my time.monotonic() calls with time.perf_counter(). I would pretty much never knowingly call time.monotonic() if I knew ahead of time it could be quantized to 16ms.

   My opinion is that the GetTickCount64() monotonic time code in CPython should be removed entirely and only the QueryPerformanceCounter() path should be used.

   I also think some of the failure checks could be removed from QueryPerformanceCounter() / QueryPerformanceFrequency(), as they're documented to never fail in modern Windows and CPython has been dropping support for older versions of Windows, but that's less of a firm opinion.
2. lunixbochs

   added

   3.8 (EOL)end of lifeend of life

   3.9 (EOL)end of lifeend of life

   3.10 (EOL)end of lifeend of life

   3.11only security fixesonly security fixes

   stdlibStandard Library Python modules in the Lib/ directoryStandard Library Python modules in the Lib/ directory

   OS-windows

   performancePerformance or resource usagePerformance or resource usage

   on Jun 6, 2021
3. 28 remaining items
4. vstinner

   added 2 commits that reference this issue on Mar 14, 2024

   pythongh-88494: Use QueryPerformanceCounter() for time.monotonic()

   2b670fc

   pythongh-88494: Use QueryPerformanceCounter() for time.monotonic()

   84632e9
5. vstinner commented on Mar 14, 2024

   

   vstinner

   on Mar 14, 2024

   Member

   More actions

   I created #116781 to use QueryPerformanceCounter() for time.monotonic().
6. vstinner commented on Mar 14, 2024

   

   vstinner

   on Mar 14, 2024

   Member

   More actions

   See also issue gh-115637.
7. vstinner

   added 4 commits that reference this issue on Mar 14, 2024

   pythongh-88494: Use QueryPerformanceCounter() for time.monotonic()

   28d2357

   gh-88494: Use QueryPerformanceCounter() for time.monotonic() (#116781)

   846ad5a

   pythongh-88494: Use GetSystemTimePreciseAsFileTime() in time.time()

   f6eec65

   pythongh-88494: Use GetSystemTimePreciseAsFileTime() in time.time()

   4ef53f4
8. vstinner commented on Mar 15, 2024

   

   vstinner

   on Mar 15, 2024

   Member

   More actions

   time.monotonic() now uses QueryPerformanceCounter() (commit) and has a resolution lower than 1 us. I measured 100 ns (10 MHz) on my Windows 11 VM, whereas before it was 15.6 ms (64 Hz): the updated clock resolution is 156,000x better :-)

   Thanks everybody for looking into this issue and providing very useful technical details about these clocks.

   Follow-up: see issue gh-63207 to use GetSystemTimePreciseAsFileTime() for time.time().
9. vstinner

   closed this as completedon Mar 15, 2024
10. github-project-automation

    moved this to Done in Date and time issues 🕰️on Mar 15, 2024
11. vstinner

    added a commit that references this issue on Mar 20, 2024

    pythongh-88494: Use QueryPerformanceCounter() for time.monotonic() (p…

    ae432c8
12. adorilson

    added a commit that references this issue on Mar 25, 2024

    pythongh-88494: Use QueryPerformanceCounter() for time.monotonic() (p…

    64bd681
13. diegorusso

    added a commit that references this issue on Apr 17, 2024

    pythongh-88494: Use QueryPerformanceCounter() for time.monotonic() (p…

    1169a0a
14. vstinner

    added a commit that references this issue on Apr 29, 2024

    pythongh-88494: Reduce CLOCK_RES to 1 ms in tests

    be949bb

Issue actions