CUDA Programming in Python

Here we give a high level introduction to using Numba, a JIT compiler for Python. Numba is a cross platform solution. Numba can generate machine code for those crucial time constrained functions. In addition, you can use Numba to generate CUDA code directly from Python code!

We're running this demonstration on the NVIDIA AGX Orin Developer Kit, the premier computer for edge computing.

00:00 Introduction
00:11 Timing a Sobel Filter
04:06 Speed Comparison CPU vs GPU
04:22 Writing CUDA in Python
07:19 Overview of how JIT works

Join this channel to get access to perks:

As an Amazon Associate I earn from qualifying purchases.

Some of these links here are affiliate links. As an Amazon Associate I earn from qualifying purchases at no extra cost to you.

[lashlarue]

Hey Jim, for your development work do you generally do it all on the Jetson-type devices or do you do it on a PC running Linux and a Nvidia video card? Great video by the way!!

[amorpheuses]

Yikes the Orin kit is nice but something I can't afford. Is there a poor man's setup you can recommend?

[Retroiu_]

Hi, great video by the way, I have a question if you let me, as far as I understand this can be done in a Jetson Nano 2GB, but there is not a clear path to follow regarding the installation and application on the jetson nano, can you lead me to one source that you know? the second question is, are there other ways to code in a Cuda environment with Python using a Jetson Nano?
thank you so much by the way, I have been following your work for a while and you have helped me a lot and saved me lots of time

[sarthakbhardwaj]

please make a video on running Small Language models or Quantised one's from hugging face on JETSON AGX xavier

[froimv]

Like for Greta meme. Also good job otherwise.