SSUG::Digital: 008 – Scalable multi-node training for AI workloads on NVIDIA DGX, Red Hat OpenShift and IBM Spectrum Scale

Nvidia and IBM did a complex proof-of-concept to demonstrate the scaling of AI workload using Nvidia DGX, Red Hat OpenShift and IBM Spectrum Scale at the example of ResNet-50 and the segmentation of images using the Audi A2D2 dataset. The project team published an IBM Redpaper with all the technical details and will present the key learnings and results.

Download slides here

Related material

• IBM Redpaper: Deployment and Usage Guide for Running AI Workloads on Red Hat OpenShift and NVIDIA DGX Systems with IBM Spectrum Scale
• Episode 6: Persistent Storage for Kubernetes and OpenShift environments

User group host: Bob Oesterlin

Speakers:

Speaker Name	Photo	Bio
Thomas Schoenemeyer		Thomas Schoenemeyer is a Senior Solution Architect and works for the NVIDIA EMEA Automotive Enterprise Team in Munich since March 2019. His area of expertise covers defining the computing architecture and infrastructure for the development of autonomous vehicles. He is helping customers to solve their technical challenges by integrating NVIDIA datacenter products such as the DGX family, the DriveSim/Constellation and NVIDIA Mellanox Networking solutions.
Gero Schmidt		Gero Schmidt is a Software Engineer at IBM Germany Research and Development GmbH in the IBM Spectrum Scale development group. He joined IBM in 2001 working at the European Storage Competence Center in Mainz, Germany, providing technical presales support for a broad range of IBM storage products with a primary focus on enterprise storage solutions, system performance, and IBM POWER systems. He participated in the product rollout of the IBM System Storage DS6000/DS8000 series, co-authored several IBM Redbooks and has been a frequent speaker at IBM international conferences. In 2015 he joined the storage research group at the IBM Almaden Research Center in California, USA, where he worked on IBM Spectrum Scale, compression of genomic data in next generation sequencing pipelines, and the development of a cloud-native backup solution for containerized applications in Kubernetes/Red Hat OpenShift. In 2018 Gero joined the Big Data & Analytics team in the IBM Spectrum Scale development group.
Simon Lorenz		Simon Lorenz is an IT Architect at IBM Research and Development in Frankfurt, Germany. He joined IBM Germany in 1993 and has held various positions within IBM and IBM Research and Development. During international assignments, he helped to improve fully automated chip factories in Asia and US. Simon joined the IBM Spectrum Scale development team in 2014 and since then worked on: Openstack Swift integration, designing and building System Health and Proactive Services solutions. Since March 2019 he is the world wide leader of the IBM Spectrum Scale Big Data and Analytics team. His role includes designing and building solutions in the area of AI such as the Data Accelerator for AI and Analytics.