Fundamentals of Accelerated Computing with CUDA Python

Please note that this workshop is limited to 30 participants due to it being run as a hands-on workshop using online resources. Unfortunately as there is no facility to close registrations after the first 30, access to the workshop on the day will be based on who accesses the resources first.

NB: In order to access the online content used in this tutorial, all participants should, prior to the start of the tutorial follow the instructions below:
- Create or log into your NVIDIA Developer Program account. This account will provide you with access to all of the DLI training materials during and after the workshop.
- Visit websocketstest.courses.nvidia.com and make sure all three test steps are checked “Yes.” This will test the ability for your system to access and deliver the training contents. If you encounter issues, try updating your browser. Note: Only Chrome and Firefox are supported.
- Check your bandwidth. 1 Mbps downstream is required and 5 Mbps is recommended. This will ensure consistent streaming of audio/video during the workshop to avoid glitches and delays.

Facilitated by: Karen Bradshaw

This tutorial teaches the fundamental tools and techniques for running GPU-accelerated Python applications using CUDA® and the NUMBA compilers. Participants will work through dozens of hands-on coding exercises, and at the end of the training, implement a new workflow to accelerate an example program originally designed for CPUs, observing impressive performance gains.

On completion of this tutorial, participants will have an understanding of the fundamental tools and techniques for GPU-accelerated Python applications with CUDA and Numba including the ability to:

GPU-accelerate NumPy ufuncs with a few lines of code
Configure code parallelization using the CUDA thread hierarchy
Write custom CUDA device kernels for maximum performance and flexibility
Use memory coalescing and on-device shared memory to increase CUDA kernel bandwidth
Generate random numbers on the GPU

Department of Computer Science, Rhodes University

Who is the facilitator?

Associate Professor in the Department of Computer Science, Rhodes University, South Africa
NVIDIA Deep Learning Institute (DLI) Certified Instructor in CUDA C/C++ and CUDA Python
PhD in Computer Science from Cambridge University, UK
Research interests: Distributed and parallel programming (including GPGPU), programming languages, simulation and modeling
For more information, see departmental website here

What does the tutorial session involve?

Scheduled as a half-day tutorial (2 x 90 min slots)
Each participant will need to supply his/her own laptop with access to the Internet. Note that no GPU is required.
The content of the tutorial is available online through the use of Jupyter Notebooks and with all practical exercises executed on a fully configured, GPU-accelerated workstation in the cloud.
After the tutorial, participants will still be able to access the material. Thus, whatever cannot be covered during the allocated conference session, can be completed individually later.
Upon successful completion of the final assessment, participants will receive an NVIDIA DLI certificate to recognize their subject matter competency.
NB: In order to access the online content used in this tutorial, all participants should, prior to the start of the tutorial follow the instructions below:
- Create or log into your NVIDIA Developer Program account. This account will provide you with access to all of the DLI training materials during and after the workshop.
- Visit websocketstest.courses.nvidia.com and make sure all three test steps are checked “Yes.” This will test the ability for your system to access and deliver the training contents. If you encounter issues, try updating your browser. Note: Only Chrome and Firefox are supported.
- Check your bandwidth. 1 Mbps downstream is required and 5 Mbps is recommended. This will ensure consistent streaming of audio/video during the workshop to avoid glitches and delays.

Who should attend?

Given that Python is often used as the coding language for many AI applications, having the ability to leverage the processing power of the GPU with more direct control than merely through the use of GPU libraries, gives developers greater freedom and flexibility in their coding.

Therefore, this tutorial is aimed at anyone developing applications in Python. By making use of the Numba (just-in-time) compiler, CPU-based applications can be easily modified to execute on a GPU with possible performance gains.

Participants should ideally:

1. be competent writing basic Python, including familiarity with variable types, loops, conditional statements, functions, and array manipulations,

2. be familiar with the NumPy Python library for numerically-focused Python; however, if they have never used NumPy, but are familiar with Python, the use of NumPy in this tutorial should be straightforward,

3. understand what vectors and matrices are, and also matrix multiplication.

Note that no previous knowledge of CUDA programming is required.

Page updated

Report abuse