Intel® Omni-Path Architecture: Speed and Scale for HPC Workloads - Intel® Chip Chat episode 553

Published: Oct. 6, 2017, 4:10 p.m.

Scott Misage, General Manager for the Intel® Omni-Path Business Unit, joins us for an update on Intel® Omni-Path Architecture (Intel® OPA). Intel OPA, part of Intel® Scalable System Framework, is a high-performance fabric enabling the responsiveness, throughput, and scalability required by today's and tomorrow's most-demanding high performance computing (HPC) workloads. In this interview, Misage talks about market uptake in Intel OPA's first year of availability, reports on some of the first HPC deployments using the Intel Xeon Scalable platform and Intel OPA, and gives a sneak peek of what Intel OPA will be talking about at SC17. For more information on Intel Omni-Path Architecture, please visit http://intel.com/omnipath. 1:11: More systems in the Top 500 than InfiniBand based on June 2017 Top500 List https://www.top500.org/list/2017/06/. 1:18: Higher percentage of systems in Top100 compared to Infiniband EDR based on June 2017 Top500 List https://www.top500.org/list/2017/06/. 2:01: Better price/performance than Infiniband: Configuration assumes a 750-node cluster, and number of switch chips required is based on a full bisectional bandwidth (FBB) Fat-Tree configuration. Intel® OPA uses one fully-populated 768-port director switch, and Mellanox EDR solution uses a combination of 648-port director switches and 36-port edge switches. Mellanox component pricing from www.kernelsoftware.com, with prices as of April 4, 2017. Compute node pricing based on Dell PowerEdge R730 server from www.dell.com, with prices as of April 4, 2017. Intel® OPA pricing based on pricing from www.kernelsoftware.com as of August 15, 2017 *Other names and brands may be claimed as property of others.