As part of the Gray Scott Thursdays webinar series, a new session will explore how to accelerate Python workloads with GPUs and modern high-performance computing tools. The session is scheduled for Thursday, 7 May 2026.
Python is popular for its simplicity and rich ecosystem, but achieving high performance in compute-heavy tasks often requires hardware accelerators such as GPUs. This webinar will guide participants through practical ways to speed up Python applications and select the right tools for their needs.
The session will explore various approaches depending on the level of integration:
- High-level frameworks such as CuNumeric, DPNP, JAX, and PyTorch, which facilitate GPU acceleration while maintaining a code structure very similar to CPU implementations.
- Intermediate tools like CuPy, which offer advanced numerical features with some modification of code.
- Low-level methods including PyCUDA and Numba, giving detailed control over GPU processes.
Additionally, the webinar will cover profiling tools designed to analyze performance and help optimize code effectively.
