ABSTRACT
It is well known in the HPC community that latency is critical and that CLOS architectures are key to scalable, low-latency, high-bandwidth systems. The critical element in the further scalability of these systems is the emergence of a robust, multi-vendor ecosystem for low-latency interconnect components with a low cost. Ethernet has the cross-market appeal, and comprises a majority of HPC systems, but is not yet seen as a performance leader. This presentation will expose the audience to new capabilities in Ethernet that make it suitable for the highest performance systems. In particular, TCP/IP/Ethernet interconnect technology now has: 1) 10 Gbps data rate 2) Extremely low latency 3) Full protocol offload and acceleration 4) Scalable, non-blocking topologies (fat trees, etc.) 5) Congestion management enhancements 6) Aggressive Ethernet-ecosystem pricing The presentation will include simulation results to show interconnect and congestion management performance in an HPC cluster from an end-to-end system perspective.
Index Terms
- Low-latency ethernet: the ubiquitous datacenter interconnect
Recommendations
A low-cost and latency bypass channel-based on-chip network
The number of cores on the chip increases rapidly; therefore, scalability is the most important design choice. Mesh-based Networks-on-Chip (NoC) are the most widely used topologies as a scalable alternative for traditional shared bus in many-core chips ...
XYZ-Randomization using TSVs for Low-Latency Energy Efficient 3D-NoCs
NOCS '17: Proceedings of the Eleventh IEEE/ACM International Symposium on Networks-on-ChipIn this paper, we propose a method to design low latency and low energy networks for 3D Network-on-Chip (3D-NoC). Recent many-core processors require low-latency interconnection networks since the increasing number of cores limits the network ...
Low latency network-on-chip router microarchitecture using request masking technique
Network-on-Chip (NoC) is fast emerging as an on-chip communication alternative for many-core System-on-Chips (SoCs). However, designing a high performance low latency NoC with low area overhead has remained a challenge. In this paper, we present a two-...
Comments