-
Notifications
You must be signed in to change notification settings - Fork 1.1k
Minutes_2021_03_09
Valentin Haenel edited this page Mar 9, 2021
·
1 revision
Attendees: kalyan, Ehsan, Todd, Jim P., Luk, Graham, Stuart, Siu, Val
NOTE: All communication is subject to the Numba Code of Conduct.
- RC3/Final status
- Reported problems:
-
https://numba.discourse.group/t/helping-test-numba-0-53-0-rc/519/22
- likely NumPy behavior specific
- non-blocking
-
https://numba.discourse.group/t/helping-test-numba-0-53-0-rc/519/22
- Reported problems:
- Roles continued
- Roles:
- no disagreement so far; will proceed to draft document for governance. (details in last week's meeting notes)
- Bigger issue to get more maintainers
- skill-set mismatch from user to maintainer
- incentives
- maintainer recognition; e.g. pandas has a page of maintainers
- "Powered By" page
- to attract SE/CS type
- ripple effect on scientific applications
- Is it advertising?
- invert problem? Should we have a "powered by sticker"?
- Numba mentorship program
- Commit to a few hours a week
- Learn how to triage (debug) Numba issues
- An experienced maintainer will be assigned to guide you
- Prior contributions may be a requirement
- or even an existing contribution
- Revisit in 4 weeks
- Roles:
- #6807 - Invalid use of Function(<ufunc 'isnan'>) with argument(s) of type(s): (int64)
- #6806 - Breaking Changes in Intel TBB 2021.1 - Numba Fails to Build
-
#6803 - ListType to numpy array issue
- Feature request for
np.array(typedlist)
- Feature request for
- #6797 - Issue when using 'not in' in numba
-
#6796 - Parallel is two times slower in LIF neuron model
- Further discussion for discourse
- #6784 - parfors invalid/outdated varmap
- #6783 - parfors lowering error from sum with internal prange (suspect SSA related!?)
- #6782 - parfors invalid result from nested reduction
- #6781 - Multiple overload method/attribute templates for same type overlap
-
#6778 - Refactor and move
create_temp_module
intonumba.tests.support
-
#6777 - CudaSupportError: Error at driver init: [100] Call to cuInit results in CUDA_ERROR_NO_DEVICE:
- Graham to suggest ctypes cudart first
- RAPIDS working with docker image setup?
-
#6808 - #3468 continued: Add support for
np.clip
-
#6804 - added runtime version to
numba -s
- #6802 - Add lineinfo flag to PTX and SASS compilation
- #6801 - ParallelAccelerator no long will convert StaticSetItem to SetItem because record arrays require StaticSetItems.
- #6800 - Make a common standard API for querying ufunc impl
- #6799 - Print Numba and llvmlite versions in sysinfo
- #6798 - CUDA: Add optional Driver API argument logging
- #6795 - CUDA: Lazily add libdevice to compilation units
- #6794 - refactor and move create_temp_module into numba.tests.support
- #6790 - Wip/pr6762 addoptionstack
- #6788 - Add support for fastmath 32-bit floating point divide
- #6775 - CUDA: Use NVVM bitcode IR interface
- #6792 - Update changelog for 0.53.0rc3
- #6791 - Preparing for 0.53.0rc3
- #6789 - Fix warnings struct ref typeguard
- #6787 - Fix typeguard test for pr6767
- #6786 - Fix gufunc kwargs support
- #6780 - Wip/pr/6767
- #6779 - Fix wrap_index type unification.
- #6776 - Fix issue with TBB interface causing warnings and parfors counting them