Enables composability of two or more multi-threaded modules by using Intel® Threading Building Blocks (Intel® TBB) library as the common work scheduler. Prevents performance degradation due to resource oversubscription without any source code change. Drop-in composable replacement for Python's Pool and ThreadPool. Switches MKL-accelerated numpy/scipy to use TBB instead of OpenMP. Additional details can be found in our [SciPy 2018 conference proceedings] (http://conference.scipy.org/proceedings/scipy2018/pdfs/anton_malakhov.pdf).
One of many Intel® accelerated Python packages and performance library runtimes available on PyPI, and as part of Intel® Distribution for Python.
For latest release updates and security notifications, please subscribe to the Intel® Distribution for Python Community forum.
