Fundamentals of Accelerated Computing with CUDA Python
mercredi 5 juin 2024 -
08:30
lundi 3 juin 2024
mardi 4 juin 2024
mercredi 5 juin 2024
08:30
Introduction - Introduction to CUDA Python with numba
-
François Courteille
Introduction - Introduction to CUDA Python with numba
François Courteille
08:30 - 10:00
Room: Jean Lascoux
Meet the instructor. Create an account at https://learn.nvidia.com/join -- 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.
10:00
Coffee Break
Coffee Break
10:00 - 10:15
Room: Jean Lascoux
10:15
Introduction to CUDA Python With Numba
-
François Courteille
Introduction to CUDA Python With Numba
François Courteille
10:15 - 12:30
Room: Jean Lascoux
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.
12:30
Lunch Break
Lunch Break
12:30 - 13:30
13:30
Custom CUDA Kernels in Python with Numba
-
François Courteille
Custom CUDA Kernels in Python with Numba
François Courteille
13:30 - 15:15
Room: Jean Lascoux
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.
15:15
Coffee Break
Coffee Break
15:15 - 15:30
Room: Jean Lascoux
15:30
Multidimensional Grids, and Shared Memory for CUDA Python with Numba
-
François Courteille
Multidimensional Grids, and Shared Memory for CUDA Python with Numba
François Courteille
15:30 - 17:15
Room: Jean Lascoux
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.
17:15
Final review
Final review
17:15 - 17:30
Room: Jean Lascoux
Review key learnings and wrap up questions. Take the workshop survey.