The Parallel Universe Magazine

Intel’s quarterly magazine helps you take your software development into the future with the latest tools, tips, and training to expand your expertise.
 

Subscribe
Issue 46, October 2021

Feature: Get Started with Habana Gaudi* for Deep-Learning Training

This issue focuses on high-performance data science and analytics, in particular, training machine-learning and deep-learning models.

Henry Gabb, editor in chief and Intel senior principal engineer, introduces how to:

  • Use the Gaudi Habana* processor unit architecture for deep learning
  • Make the most of Intel® optimizations for working with AI in the cloud
  • Improve the scalability of training deep-learning recommender systems using the results of a recent collaboration with Facebook*
Download the PDF (6.4 MB)
Contents
Letter from the Editor: Let’s Talk about High-Performance Data Analytics

Henry A. Gabb, senior principal engineer, Intel

Getting Started with Habana® Gaudi® for Deep Learning Training

A New Hardware and Software Stack for Deep Learning

Speeding Up the Databricks Runtime for Machine Learning

Intel-Optimized Artificial Intelligence in the Cloud

A Novel Scale-Out Training Solution for Deep Learning Recommender Systems

Demonstrating Better Parallel Scaling in the MLPerf Benchmark

Cost Matters: On the Importance of Cost-Aware Hyperparameter Optimization

Accounting for the Cost Metric that Actually Matters

Katana's High-Performance Graph Analytics Library

Python Programmers Have a New Option for Graph Analytics

Accelerate R Code with Intel® oneAPI Math Kernel Library

Improve R Performance without Modifying Code

Optimizing the Maxloc Operation Using Intel® AVX-512 Instructions

A Guide to Vectorizing this Common Reduction Operation