Using Python to compare methods of running data parallel operation on the CPU and GPU
0:00 Introduction
2:50 PC description
3:48 Data Parallel vs Task Parallel
6:50 Using anaconda for Python Environments
8:41 How system IO (printing) slows down processing
12:30 Running Python on the GPU (matrix operations)
23:25 When to use CPU instead of GPU
28:30 GPU array creation and memory use
33:15 Effect of Data type choice on speed and memory use
35:47 Vectorized operations
41:54 Reading GPU specs with numba -s
44:52 Forcing Out of Memory error on GPU
46:50 Conclusion
Useful links for setting up to test data parallel python code:
https://www.anaconda.com/
https://docs.cupy.dev/en/stable/index...
https://numba.pydata.org/
https://numpy.org/
https://intelpython.github.io/dpnp/