Numba 0.55.0 and llvmlite 0.38.0 are now available.
NOTE: Due to NumPy CVE-2021-33430 this release has bypassed the usual release process so as to promptly provide a Numba release that supports NumPy 1.21. A single release candidate (RC1) was made and a few issues were reported, these are summarised as follows and will be fixed in a subsequent 0.55.1 release.
Known issues with this release:
- Incorrect result copying array-typed field of structured array (Incorrect result copying array-typed field of structured array in 0.55.0rc1 · Issue #7693 · numba/numba · GitHub).
- Two issues in DebugInfo generation (DWARF generation failing for OmittedArgs · Issue #7726 · numba/numba · GitHub and "Unexpected" indentation can break DI emission · Issue #7730 · numba/numba · GitHub).
- Compilation failure for
hashof floating point values on 32 bit Windows when using Python 3.10 (Tests using float hashing fail on win32+Py3.10 for 0.55rc1 · Issue #7713 · numba/numba · GitHub).
Highlights of core changes:
- Support for Python 3.10
- Support for NumPy 1.21
- Experimental support for
- A lot of DWARF improvements (for debugging with gdb!)
- Support for NVIDIA’s CUDA Python bindings.
- Some more NumPy functions are now supported.
For details, please refer to the change logs at:
Wheels are in the usual place and conda packages are in the
conda install -c numba numba=0.55.0 conda install -c numba llvmlite=0.38.0
There are no conda packages for:
- linux-ppc64le with Python 3.10, there’s an issue with the compiler tool-chain that is preventing compilation of shared libraries.
- Windows 32 bit with Python 3.10, the bug noted above (Tests using float hashing fail on win32+Py3.10 for 0.55rc1 · Issue #7713 · numba/numba · GitHub) is preventing the build from passing testing, as a result the packages will not ship, this build will be available for 0.55.1.
The Numba team