OpenPOWER and Open Source Innovation Shine at IBM TechU Rome
Published on Wednesday 7 November 2018
By Florin Manaila, Senior IT Architect and Inventor, IBM Systems Hardware Europe
The IBM Technical University in Rome was the perfect place to gather a large audience from around the globe on topics including AIX, Cognitive, Linux on Power, Power Systems Storage and SDI.
TechU Rome 2018 event offered unique initiatives that attracted many technical persons during the week, including a PowerAI Meetup that discussed AI deep learning, a PowerAI Discussion Group on OpenPOWER hardware as well as a PowerAI Lounge where presentations on IBM OpenPower Systems Interact with DEMO’s and other IBM and BP experts.
TechU Rome 2018 was a great opportunity to watch an IBM BluePrint for Deep Learning in Edge device and how PowerAI models can be successfully used on small, accelerated devices by embedding GPUs such as NVIDIA Jetson TX2..
Across all sessions was a pattern: the value of the open platform and innovation. IBM OpenPower and Open Source innovation was presented on many topics, summaries of the sessions can be found below:
OpenCAPI: Next generation of acceleration for the cognitive area
By Myron Slota
Open Coherent Accelerator Processor Interface (OpenCAPI) is a new industry standard device interface. OpenCAPI enables the development of host-agnostic devices which can coherently connect to any host platform which supports the OpenCAPI standard, providing the device with the capability to coherently cache host memory to facilitate accelerator execution. This session describes where we think acceleration is going in the industry, the disruptive technologies in the acceleration space, and where IBM Power Systems will be participating in acceleration.
IBM AC922 Deep Learning System
By Florin Manila
GPU Deep Learning is the foundation for the fourth industrial revolution driven by AI. During this session will be presented the IBM AC922 architecture and design for deep learning workloads. As a fundamental block for Distributed Deep Learning Architecture various sizing aspects will be presented together with related PowerAI software implications.
Open Source and Power Systems
By Andrew Laidlaw
A look at a selection of the Open Source Communities and Projects that are relevant to Power Systems, on both the hardware and software side of innovation. As well as the obvious examples of the OpenPOWER Foundation and OpenCAPI Consortium, we will look at how Open Source is influencing the software landscape, and the Power Systems strategy. This will include looking at the Deep Learning space, both SQL and NoSQL databases, and deployment technologies like OpenStack and Kubernetes.
Linux On Power: Trends and Directions
By Steven Roberson
This session will present the roadmap for the various Linux distributions on Power, including OS, OpenStack and Container strategy. We will also discuss the capabilities of the Power 9 processor and the architecture that the distributions exploit. In addition, we will cover the hardware roadmap of all Power servers built for Linux workloads.
FPGA with CAPI : an alternative to accelerate your apps
By Alexandre Castellane
This lecture will help you understand what FPGA hardware acceleration provides and when it can be used in complement or replacement of GPU. SNAP framework provides software engineers with a mean to use this technology in a snap ! The unique advantages of POWER technology including CAPI / OpenCAPI technology coupled to SNAP framework will be presented. Details of what memory coherency and low latency associated to FPGA bring to you will be explored through very simple examples.
Architecting the Future: Innovating with Data, AI and Cloud
By Thomas Harrer
The Architecting the Future track has been representing a consistent IBM track on how the technology can change our life’s and industries such healthcare, automotive by adopting AI. An efforts show business value behind the technology, for architecture, client use cases, advantages, and why IBM is the best partner to innovate with.
All the presentations are available here.