StingraySoftware
HENzsearch and tbb library
When using HENzsearch a warning appears saying TBB threading layer is disabled. Is this an important issue ?
HENzsearch -f .1 -F 2. -N 2 --fast ULX_xmm_epn_det04_ev.nc 0%| | 0/17756 [00:00<?, ?it/s]/home/gianluca/stingray/lib64/python3.6/site-packages/numba/np/ufunc/parallel.py:365: NumbaWarning: The TBB threading layer requires TBB version 2019.5 or later i.e., TBB_INTERFACE_VERSION >= 11005. Found TBB_INTERFACE_VERSION = 10002. The TBB threading layer is disabled. warnings.warn(problem) 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 17756/17756 [27:44<00:00, 10.67it/s]
I tried to check whether tbb is present on my system:
pip install tbb Requirement already satisfied: tbb in /data/gianluca/stingray/lib/python3.6/site-packages (2021.2.0)
@stuartarchibald looking at a previous Numba issue, the solution should be a tbb update. However, here the installed version (2021.2.0) seems more recent than needed (2019.5 or later). Any ideas? Thanks in advance
@giallixx have you tried with a more recent python version, e.g. 3.8?
TBB changed its ABI in the 2021 series. Numba 0.53.x (latest) needs TBB 2019.5-2020.x, the upcoming Numba 0.54 uses TBB 2021 (the conda packages encode this). IIRC there's also a general problem with detection of pip installed TBB in that Numba sometimes can't see it because it could well not be on a path the dynamic link resolver knows about.
python version 3.6 is the latest available version on the linux distribution I am currently using.