Minutes_2021_06_15
Valentin Haenel edited this page Jun 16, 2021
·
1 revision
Attendees: Siu Kwan Lam, Todd A Anderson, Brandon Willard, Graham Markall, Stuart, Valentin Haenel
NOTE: All communication is subject to the Numba Code of Conduct.
- Discuss w/ Brandon on broadcasting support related to
- Moving off gitter
- https://numba.discourse.group/t/request-for-thoughts-feedback-on-moving-off-gitter-potentially-to-slack/751
- other clients
- zulip
- further reading: https://zulip.com/for/open-source/
- ms team
- pumble.com
- zulip
- also looked at
- Discord
- but lack threads currently
- Discord
- Brandon talks about Aesara performance
- issue: https://github.com/aesara-devs/aesara/issues/404
- numba performance is better than generated C
- function call overhead is lower
- #7112 - NumPy Ufuncs on scalar values don't work in CUDA
- #7111 - Using @overload with types.Callable fails to find compiled implementation at Lowering
- #7110 - numpy.delete optional argument 'axis' support
- #7109 - Second update to nonlocal variable is ignored
- #7108 - Mutating a typed dict whilst iterating is not handled correctly
- #7105 - Array returned from prange loop is wrong unless that array is read before the function returns.
- #7104 - CUDA on WSL2 Support
- #7103 - GDB doesn't show the values of variables when it hits breakpoint the first time
- #7101 - segmentation fault when calling np.all in ipython
- #7099 - NamedTuple Optional np.ndarray field type does not compile
- #7095 - numba build fails with TBB
- #7093 - CUDA: Make typing and target contexts singletons
- #7091 - Numba causes crash when using python climate-indices package in terminal
- #7114 - CUDA: Deprecate eager deprecation of device functions
- #7107 - CUDA: Fold the device dispatcher into the kernel dispatcher
- #7106 - Use correct CUDA driver library on WSL
- #7100 - Replace deprecated call to cuDeviceComputeCapability
- #7098 - [TESTING ONLY] Temp/wip/dispstack alt2
- #7096 - Ddof support for nanvar nanstd