Releases: plasma-umass/scalene
Releases · plasma-umass/scalene
UI and CLI enhancements, improved multiprocessing/module support, and Windows bugfixes
What's Changed
- Now generates HTML instead of using a web server by @emeryberger in #477
- Clearer command-line parameters (
--cpu
,--gpu
, and--memory
) by @emeryberger in #477 - Improved multiprocessing and module support by @RuRo in #484
- Fixes several issues running Scalene on Windows by @emeryberger
New Contributors
Full Changelog: v1.5.14...v1.5.15
Maintenance release: various bug fixes
Changes in this release:
Bug fixes:
- Disabled GPU profiling for very old NVIDIA platforms that don't support profiling utilization and/or memory consumption; this used to lead to failures
- More graceful handling of non source files (which could lead to failures)
- Uses correct path regardless of changes to the working directory
UI:
- forced profiles in Jupyter cells to consume 100% of the available width
- minor usability fix for Jupyter, fixing an issue with
%scrun
Full Changelog: v1.5.13...v1.5.14
Disable GPU profiling for Apple, since it is unreliable
Changes in this release:
- Disabled Apple GPU profiling for now since it is unreliable (the Apple GPU does not record activity on a per-process basis).
Full Changelog: v1.5.12...v1.5.13
Improved memory reporting
Changes in this release:
- Corrected some memory attribution errors (primarily relating to calculating average memory per line/function, as well as a race condition).
Full Changelog: v1.5.11...v1.5.12
Maintenance release
Changes in this release:
- Fixed pre-built MacOS Universal distributions not including M1 support;
- Fixed Scalene's assumption it would remain in the same directory, which caused problems when the profiled program did a
chdir
.
Full Changelog: v1.5.10...v1.5.11
Maintenance release: fixes, warnings, and changes for Conda
Changes in this release:
- Fixed a reference counting issue that could lead to failure (73c848b).
- Increased an internal buffer size to ensure safe handling when Scalene is accessing very long directory / pathnames.
- Added support for profiling applications that themselves use
LD_PRELOAD
(fixing #418). - Improved warning message for the current lack of support for
multiprocessing
on Windows (addressing #416). - Other changes to enable building on Conda (conda-forge/staged-recipes#18747).
Full Changelog: v1.5.9.1...v1.5.10
Increased profiling accuracy, Python 3.11 support
Changes in this release:
- increased accuracy of time attribution to specific lines for CPU & GPU profiling (also reduces memory consumption)
- increased accuracy of memory attribution to specific liens
- added per-process GPU accounting for NVIDIA, which can dramatically increase accuracy when profiling on shared GPUs
- added support for Python 3.11
- documented the command-line option to force Scalene to ignore options after that point (
---
)
Full Changelog: v1.5.8...v1.5.9.1
Increased profiling accuracy, Python 3.11 support
Changes in this release:
- increased accuracy of time attribution to specific lines for CPU & GPU profiling (also reduces memory consumption)
- increased accuracy of memory attribution to specific liens
- added per-process GPU accounting for NVIDIA, which can dramatically increase accuracy when profiling on shared GPUs
- added support for Python 3.11
- documented the command-line option to force Scalene to ignore options after that point (
---
)
Full Changelog: v1.5.8...v1.5.9
Fixed GUI display issues
Changes in this release:
- fixed missing GUI files from Linux wheels;
- fixed some issues launching browser to display GUI results;
Minor UI update: memory activity now shown as pies instead of numbers
What's Changed
UI improvements:
- Memory activity now shown as pies instead of numbers
Compatibility:
- Working towards conda builds.
Full Changelog: v1.5.6...v1.5.7