NVIDIA Deep Learning Institute, together with Swinburne University of Technology and Astronomy Data & Computing Services (ADACS), is hosting a workshop of CUDA and OpenACC development on GPUs - comprising of lectures and hands-on labs, exclusively for vertifiable academic students, staff and researchers.

In this workshop, you will start with the basic programming skills of CUDA and OpenACC and quickly move on to learning how to solve real-world problems using CUDA and OpenACC.

Attendees must bring their own laptops and power supply.Connectivity will be available through the University Wi-Fi.

You will learn to:

• Understand CUDA programming skills
• Parallelize matrix multiply algorithm with CUDA
• Learn data management with Unified Memory of GPU
• Understand OpenACC programming skills
  

Agenda

09:45 Registration

10:00 Lecture: CUDA Programming Skills

11:00 Hands-on lab: Accelerating Applications with CUDA C/C++

12:30 Lunch Break

13:30 Lecture: OpenACC Programming Skills

14:30 Afternoon Break

14:45 Hands-on lab: OpenACC - 2X in 4 Steps

16:30 Running GPU jobs on Swinburne HPC Clusters (Swinburne University of Technology)

17:00 End

Content Level: Beginner

Pre-requisite: Basic C/C++ understanding can be helpful for some exercises.

IMPORTANT: To reserve your seat, you MUST register with a valid university email address and follow these pre-workshop instructions.

• You must bring your own laptop, charger and adaptor (if needed) to this workshop.
• Please note that access to the same email address used to register for the event on Eventbrite, will be required for the Qwiklabs account registration on the event day - which is used to run the hands-on labs. Please ensure you use only your university email address.
  

Training Syllabus

GPU Lecture 1: CUDA Programming Skills

This lecture will explain the GPU architecture and teach basic skills of CUDA programming, such as how to write CUDA kernels, how to utilize GPU threads, and how to optimize GPU memory access, etc. After this lecture, you will be able to understand how to parallelize your sequential code in CUDA, get ready to write and optimize your CUDA programs.

GPU Lab 1: Accelerating Applications with CUDA C/C++

Learn how to accelerate your C/C++ application using CUDA to harness the massively parallel power of NVIDIA GPUs. In 90 minutes, you will work through seven exercises, including:

• Hello Parallelism!
• Accelerate the simple SAXPY algorithm
• Accelerate a basic Matrix Multiply algorithm with CUDA
• Error checking GPU code
• Querying GPU Devices for capabilities
• Data management with Unified Memory
• A case study implementing most of the above
  

GPU Lecture 2: OpenACC Programming Skills

GPU Lab 2: OpenACC - 2X in 4 Steps

Learn how to accelerate your C/C++ or Fortran application using OpenACC to harness the massively parallel power of NVIDIA GPUs. OpenACC is a directive based approach to computing where you provide compiler hints to accelerate your code, instead of writing the accelerator code yourself. In 90 minutes, you will experience a four-step process for accelerating applications using OpenACC:

1. Characterize and profile your application
2. Add compute directives
3. Add directives to optimize data movement
4. Optimize your application using kernel scheduling

Running GPU jobs on Swinburne HPC Clusters (Swinburne University of Technology)

This session will explain how to connect to the HPC cluster and P100s GPUs based in Swinburne University of Technology.

NVIDIA Instructor:

Maggie Zhang

Maggie Zhang is a Solutions Architect in the areas of HPC/Deep Learning at NVIDIA, Melbourne. Before joining NVIDIA, she worked as a postdoctoral researcher in Lero (The Irish Software Research Center), Ireland and a research fellow in National University of Defense Technology, China. Her research areas are GPU/CPU heterogeneous computing, compiler optimization, computer architecture, deep learning. She got her PhD in Computer Science & Engineering from the University of New South Wales in 2013.