NVIDIA Workshops on Deep Learning, Pascal, CUDA & OpenACC

The YCRC and NVIDIA invite you to attend one or two days of in-depth workshop lectures covering a variety of topics including the latest NVIDIA Volta GPU architecture news and Pascal updates, hands on exposure to contemporary Deep Learning frameworks as well as NVIDIA CUDA and OpenACC.

If you would like to attend both days, you must register twice (once for each day).

Day 1- Tuesday, July 25th:

• 08:30 am - 09:00 am - Introductions & NVIDIA Updates by Barton Fiske & Brad Palmer
• 9:00 am - 09:30 am - GTC 2017 Highlights by Brad Palmer
• 9:30 am - 10:00 am - GPU Acceleration for HPC and Deep Learning
• 10:00 am - 10:30 am - Break / refreshments provided
• 10:30 am - 12:00 pm - NVIDIA GPU Architecture and CUDA Deep
• 12:00 pm - 01:00 pm - Lunch to be provided (Panera menu, vegetarian options included)
• 01:00 pm - 02:30 pm - OpenACC Tutorial - Mathew Colgrove (remote)
• 02:30 pm - 03:00 pm - Break
• 03:00 pm - 04:30 pm - OpenACC Tutorial Continued – Mathew Colgrove (remote)
• 04:30 pm - 05:00 pm - General Q&A / Meet & Greet – All

Day 2- Wednesday, July 26th:

• 08:30 am - 09:00 am - Introductions & Overview of NVIDIA AI Platform
• 09:00 am - 10:00 am - Image Classification with DIGITS
• 10:00 am - 10:30 am - Break / refreshments provided
• 10:30 am - 12:00 pm – Image Classification with DIGITS (continued)
• 12:00 pm - 01:00 pm - Lunch to be provided (Panera menu, vegetarian options included)
• 01:00 pm - 02:45 pm - Object Detection with DIGITS
• 02:45 pm - 03:00 pm – Break
• 03:00 pm - 04:45 pm - Deep Learning for Image Segmentation with TensorFlow Lab
• 04:45 pm - 05:00 pm - Wrap up summary, General Q&A - All

Detailed Session Descriptions- Day 1:

Introductions and NVIDIA Updates

A brisk overview of the agenda for the two days as well as latest/greatest breaking news on NVIDIA and topics relevant to the workshop material.

GPU Acceleration for HPC and Deep Learning

This session will explain why GPU acceleration is important and review the different ways you can take advantage of GPU acceleration. GPU programming concepts will also be introduced. This session will include information on deep learning with GPUs and NVIDIA’s Deep Learning SDK

NVIDIA GPU Architecture and CUDA Deep Dive (including CUDA9)

This session will take a deeper look at NVIDIA’s current “Pascal” GPU architecture and features added to the CUDA9 toolkit and libraries along with a peek at the upcoming “Volta” GPU architecture and CUDA9 toolkit and libraries.

OpenACC Tutorial (led remotely by Mathew Colgrove)

NVIDIA GPUs are the world’s fastest and most efficient accelerators delivering world record scientific application performance. NVIDIA CUDA is the most pervasive parallel computing model, used by over 370 scientific applications and over 150,000 developers worldwide.

This workshop will focus on introducing scientific computing and programming concepts utilizing NVIDIA GPUs to accelerate applications. The workshop will introduce programming techniques using OpenACC paradigms as well as optimization, profiling, and methods for GPU programming.

The workshop will cover:

• High Level Overview of GPU Architecture
• Introduction to GPU programming with OpenACC
• Instructore led demos of the OpenACC code optimization process

Detailed Session Descriptions- Day 2:

Image Classification with DIGITS

Deep learning is giving machines near human levels of visual recognition capabilities and disrupting many applications by replacing hand-coded software with predictive models learned directly from data. This lab introduces the machine learning workflow and provides hands-on experience with using deep neural networks (DNN) to solve a real-world image classification problem. You will walk through the process of data preparation, model definition, model training and troubleshooting, validation testing and strategies for improving model performance. You will also see the benefits of GPU acceleration in the model training process. On completion of this lab you will have the knowledge to use NVIDIA DIGITS to train a DNN on your own image classification dataset.

Object Detection with DIGITS

This lab explores three approaches to identify a specific feature within an image. Each approach is measured in relation to three metrics: model training time, model accuracy and speed of detection during deployment. On completion of this lab, you will understand the merits of each approach and learn how to detect objects using neural networks trained on NVIDIA DIGITS on real-world datasets.

Deep Learning for Image Segmentation with TensorFlow

There are a variety of important applications that need to go beyond detecting individual objects within an image and instead segment the image into spatial regions of interest. For example, in medical imagery analysis it is often important to separate the pixels corresponding to different types of tissue, blood or abnormal cells so that we can isolate a particular organ. In this lab we will use the TensorFlow deep learning framework to train and evaluate an image segmentation network using a medical imagery dataset.

General guidelines/recommendations:

• There will be a wait list for this event. If we receive any cancelations we will continue to monitor waiting list requests. Check back to the EventBrite site for any updates and if you know you cannot attend, please cancel your tickets if possible.
• Participants should bring a laptop with an SSH client already installed. (PuTTY is recommended for PC Users, Mac & Linux users can use the built in "Terminal" application). Power and wireless connectivity will be provided.
• A GPU in the laptop is not required. Hands on sessions make use of cloud based GPU services available through QwikLabs Participants are encouraged to join QwikLabs and create a user account prior to the event.
• Basic Linux desktop and command line familiarity including use of a standard file editor such as VIM or Emacs.
• Familiarity with software development tools and concepts: compiling, linking and using GNUMake.
• Rudimentary programming experience in C/C++ (memory management using malloc/free, using pointers, etc)