lecture-python-programming.myst
lecture-python-programming.myst copied to clipboard
Add a Lecture on Numba Cuda
Add a lecture based on the numba_cuda notebook written by @jstac with an introduction to basic concepts such as kernel declaration, kernel invocation, and other terminologies on GPU programming.
Thanks @HumphreyYang . This application can be used as a running example throughout our parallelization lectures: https://github.com/QuantEcon/cbc_workshops/blob/main/day_4/european_option_jax.ipynb
Although it's not published yet, that example appears in the scipy lecture and then in the first parallelization lecture. We can then show how to solve it using numba cuda, and finally using JAX.
(These thoughts probably need to be broken out into separate issues but I'll leave them here for now.)
- [ ] add GPU support for lecture series when this is written.
There may be some materials / examples in https://github.com/numba/nvidia-cuda-tutorial that will also serve as useful components of the lecture - feel free to use parts of it.
thanks @gmarkall -- those tutorials look great. Thank you!