1 / 4

Hardware and Architecture

Hardware and Architecture. Co-Leads: Ken Alvin (SNL), Travis Humble (ORNL), Katie Schuman (ORNL) Science Writer: Elizabeth Rosenthal + 25 Participants (available in spreadsheet). AI Application Requirements for Hardware. System of systems architectures from edge computing to data centers

rhamm
Download Presentation

Hardware and Architecture

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Hardware and Architecture Co-Leads: Ken Alvin (SNL), Travis Humble (ORNL), Katie Schuman (ORNL) Science Writer: Elizabeth Rosenthal + 25 Participants (available in spreadsheet)

  2. AI Application Requirements for Hardware • System of systems architectures from edge computing to data centers • Complex data-driven workflows • Energy and environmental constraints on computing, particularly at the edge • AI systems must directly integrate with sensors and control elements and survive in their operational environments • Data reduction and data movement is key to most workflows • Use of data and AI for control of systems and experiments • Energy and performance requirements drive custom, heterogeneous, and novel computational devices • Domains have specific/unique workflow requirements, driving the need for co-design tools to architect systems • AI use cases within existing HPC and data centers

  3. Research Opportunities • Capturing and characterization of workflow requirements as exemplar workflows, datasets, etc. • Adaptive and fast-changing • Co-design tools, benchmarks, and metrics • Design of distributed, adaptable system of systems architectures • Real-time and quality of service • Development, implementation, and integration of new edge computing devices with flexibility • Power constraints, radiation tolerant, data reduction, etc. • Data movement and access • Bandwidth, latency, reliability, etc. • Data center integration of specialized hardware • Design for new workflows • Scalability

  4. Enabling Technologies for Hardware Research • Co-design tools • Wide area network or system of systems simulations • Architectural simulators • Proxy workflows • Neural architecture search (meta-learning) • Specialized hardware for AI • Examples: Neuromorphic, quantum, AI accelerators, photonics, FPGAs, ASICs, etc. • Testbeds and hardware evaluation labs • In the lab hardware integrated with example sensors from edge environments • Integration of heterogeneous components

More Related