Título del curso: Fundamentals of Accelerated Computing with CUDA Python.
El curso será online y tendrá lugar los días 30 y 31 de marzo de 2022, en horario de mañana de 10:00h a 14:00h.
Con anterioridad al inicio del curso se enviarán los detalles necesarios para la conexión a los asistentes.
Dirigido a usuarios del ordenador FinisTerrae procedentes de entidades públicas, con el objetivo de dar a conocer las posibilidades de uso y programación bajo python de las tarjetas NVIDIA A100 presentes en el FinisTerrae.
Este taller será impartido en inglés por el NVIDIA Deep Learning Institute usando su infraestructura.
Sesión 1 30 marzo miércoles 10:00 – 14:00
Introduction (30 mins)
- Meet the instructor.
- Create an account at https://courses.nvidia.com/join
Introduction to CUDA Python with Numba (120 mins)
- Begin working with the Numba compiler and CUDA programming in Python.
- Use Numba decorators to GPU-accelerate numerical Python functions.
- Optimize host-to-device and device-to-host memory transfers.
Custom CUDA Kernels in Python with Numba (60 mins)
- Learn CUDA’s parallel thread hierarchy and how to extend parallel program possibilities.
- Launch massively parallel custom CUDA kernels on the GPU.
- Utilize CUDA atomic operations to avoid race conditions during parallel execution.
Sesión 2 31 marzo jueves 10:00 – 14:00
Custom CUDA Kernels in Python with Numba – continued (60 mins)
- Learn CUDA’s parallel thread hierarchy and how to extend parallel program possibilities.
- Launch massively parallel custom CUDA kernels on the GPU.
- Utilize CUDA atomic operations to avoid race conditions during parallel execution.
Multidimensional Grids, and Shared Memory for CUDA Python with Numba (120 mins)
- Learn multidimensional grid creation and how to work in parallel on 2D matrices.
- Leverage on-device shared memory to promote memory coalescing while reshaping 2D matrices.
Final Review (30 mins)
- Review key learnings and wrap up questions.
- Complete the assessment to earn a certificate.
- Take the workshop survey.