BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T000713Z
LOCATION:503-504
DTSTART;TZID=America/Denver:20231115T113000
DTEND;TZID=America/Denver:20231115T120000
UID:submissions.supercomputing.org_SC23_sess251_exforum112@linklings.com
SUMMARY:Ethernet-Based Interconnect, the Critical Crossroads for HPC and A
 I Networking at Scale
DESCRIPTION:Eric Eppe (Eviden)\n\nThe convergence of HPC and AI entails an
  explosion of performance of the number of nodes/cores, data volume, and d
 ata movement. In the coming years, the deployment of AI networks, ranging 
 from rack-scale to datacenter-scale, is set to accelerate, necessitating t
 he evolution of networking technology to effectively support and accommoda
 te these needs. \n\nEviden Vision and solution: \nThe existing HPC & AI ne
 tworking choices have their virtues to continuously support the simulation
  evolution. However, depending on the specific requirements, e.g. workload
  characteristics, performance considerations, and budget constraints, cust
 omers should have the flexibility and choices to an open, interoperable, h
 igh performance, full-communications stack architecture to meet the growin
 g network demands at scale, whether on-premises or in the cloud. \n\nAs on
 e of the founding members of the UEC (Ultra Ethernet Consortium), Eviden s
 upports an Ethernet-based, multivendor interoperable, scalable, and cost-e
 ffective high-performance networking for HPC and AI workload, focusing on:
 \n\n•	Performance and reliability:  extremely low latency for efficient co
 mmunication between nodes for both HPC and AI workloads, supporting massiv
 e data transfer or high-throughput inter-node communication. Effectively m
 anage packet processing, network congestion, and message handling protocol
 s, enabling low-latency and substantial data transfers with RDMA over Ethe
 rnet, ensuring data integrity. \n\n•	Interoperability and Simplicity: Leve
 raging the ubiquity, mature ecosystem, vendor support, and interoperabilit
 y with various software and toolsets, Ethernet-based next-gen networking f
 abric offers seamless integration with popular operating systems, cluster 
 management software, storage solutions, and distributed file systems commo
 nly used in data centers. \n\n•	Scalability and cost-effectiveness: the af
 fordability of Ethernet hardware, software and the widespread adoption in 
 commercial networks makes it a cost-effective and easy-to-scale networking
  choices. \n\nWith the new workload scale and complexity, optimizing the i
 nterconnect networking for HPC and AI systems become a major contributor t
 o global performance. Ethernet-based high-performance networking technolog
 y provides a new avenue to tackle technical and economic challenges.\n\nTa
 g: Accelerators, Artificial Intelligence/Machine Learning, Architecture an
 d Networks, Hardware Technologies\n\nRegistration Category: Tech Program R
 eg Pass, Exhibits Reg Pass\n\nSession Chair: Nathan Hanford (Lawrence Live
 rmore National Laboratory (LLNL))\n\n
END:VEVENT
END:VCALENDAR