On November 21, 2025, at 2:00 PM (WET), the EPICURE project will host the webinar “CPU-based Threading Models Application in Python for HPC Computing”, presented by Veselin Kolev, from the Discoverer Supercomputer team (Bulgaria).
The session will explore three complementary approaches to overcome Python’s Global Interpreter Lock (GIL) limitation, enabling higher performance in scientific and HPC applications:
- Using compiled C/C++/Fortran code with OpenMP or TBB for maximum performance
- Applying PyOMP with Numba, for HPC users who prefer Python development
- Advanced LLVM integration with GIL-disabled Python for professional developers
Each method offers performance gains ranging from 6x to over 10x compared to sequential Python execution, with skill requirements from basic Python knowledge to advanced LLVM IR expertise.
The webinar is designed for HPC practitioners, researchers, and developers aiming to implement efficient parallel computing in Python applications.