Webinar - Python GPU computing (Cunumeric, cuPy, JAX, Pytorch)

How to use GPU to speed up computing with Python ? In this webinar we will study possibles senarios to guide you to the library which fit your needs. We will see basic principle of each of them, and the Gray Scott summer school will give a deeper vision of their use. We will present them by level of integration:
- for high level: Nvidia CuNumeric, Intel DPNP, JAX, and PyTorch allow to keep practically the same code as the CPU version
- for intermediate level: CuPy library provides access to advanced digital functionnalities, but needs specific code for the GPU
- for low level: PyCUDA and Numba for GPU provide basic functionnalities on tables.

We will also present associated profiling tools.

Webinar by Alice Faure (LUPM / IN2P3 / CNRS), Nabil Garroum (LPNHE / IN2P3 / CNRS) and Jean-Marc Colley (LPNHE / IN2P3 / CNRS)

The "Gray Scott Thursdays" are a special series of 17 weekly webinars preparing for the Gray Scott Summer School, covering topics such as:
⚙️ Unit testing, CPU & GPU architectures
💡 Computing precision, memory allocation
🌍 Programming with modern C++, Rust, Fortran, Python, Sycl, EVE, CUDA... and more!

Follow us to get the latest news about the #GrayScottSchool2025