ROCm and Hopsworks for end-to-end deep learning pipelines

Jim Dowling (Logical Clocks), Ajit Mathews (AMD)

16:50–17:30 Thursday, 17 October 2019

Location: King's Suite - Sandringham

Implementing AI

Secondary topics: Deep Learning tools, Hardware, Machine Learning tools

Average rating:

(5.00, 1 rating)

Download slides (PDF)

Who is this presentation for?

Data scientists, data architects, data engineers, and CTOs

Level

Intermediate

Description

The Radeon open ecosystem (ROCm) is an open source software foundation for GPU computing on Linux. ROCm supports TensorFlow and PyTorch using MIOpen, a library of highly optimized GPU routines for deep learning.

Jim Dowling and Ajit Mathews outline how Hopsworks, an open source platform for machine learning infrastructure, enables the training and operation of deep learning models on ROCm in horizontally scalable end-to-end machine learning pipelines. He presents performance benchmarks for ROCm on new GPU hardware (AMD MI50, MI60 GPUs) and shows you how Hopsworks can enable distributed deep learning with both ROCm and Cuda on both TensorFlow and PyTorch. You’ll see a live demonstration of training and inference for an end-to-end machine learning pipeline written in a number of Jupyter notebooks orchestrated by Airflow.

Prerequisite knowledge

A basic understanding of Python and either TensorFlow or PyTorch programs

What you'll learn

Learn that ROCm is a viable, high-performance platform for deep learning and Hopsworks enables the development and operation of horizontally scalable deep learning pipelines using ROCm

Jim Dowling

Logical Clocks

Jim Dowling is the CEO of Logical Clocks, an associate professor at KTH Royal Institute of Technology in Stockholm, and lead architect of Hopsworks, an open source data and AI platform. He’s a regular speaker at big data industry conferences. He holds a PhD in distributed systems from Trinity College Dublin.

Website

Ajit Mathews

AMD

Ajit Mathews is the vice president of machine learning software engineering at AMD, where he’s the engineering leader responsible for the design and development of Radeon Open Compute (ROCm) machine intelligence software spanning deep learning frameworks, compilers, language runtimes, libraries and Linux compute kernel. Ajit is also responsible for the machine learning software road map and strategy. He’s passionate about distributed machine learning and high-performance computing. Ajit holds a master’s degree in computer science and an MBA from Kellogg.