Innovium outlined a series of enhancements for its TERALYNX architecture for enabling top-to-bottom, 1T to 51.2T+ performance scalability along with programmability, telemetry, low latency and large on-chip buffers.
Innovium’s Enhanced TERALYNX architecture consists of:
- Programmable feature-rich INNOFLEX™ Forwarding Pipeline: A programmable forwarding pipeline architecture that delivers comprehensive data center feature set, including rich tunneling support, with deterministic low latency, and ability to support new protocols through standards-based programming.
- Highly Scalable TERASCALE Fabric: A scalable and low-latency switch & buffer fabric architecture that delivers comprehensive QoS, superior lossless traffic management, large smart packet buffers and scales from 1 to 51.2Tbps+ performance. TERASCALE deliver up to 2x performance and buffering in the same silicon area vs alternatives.
- Advanced FLASHLIGHT Telemetry & Analytics: A ground-up comprehensive telemetry architecture that delivers unmatched real-time visibility and actionable granular network analytics to troubleshoot and resolve network congestion problems helping customers move towards higher-quality, autonomous networks.
- An optimized R&D flow that delivers lower silicon area along with lower latency vs other architectures.
Innovium said these breakthrough innovations span a variety of areas including data-structures, algorithms, ground-up architecture and R&D flow and deliver up to 2x capabilities and performance in the same silicon area vs alternatives. This technology has been silicon proven as part of the production-ready TERALYNX 7, which delivered the best feature-set 12.8T switch silicon, with lowest-latency in 16nm and smallest silicon area. The enhanced architecture further optimizes silicon area and data-center features to expand the performance range and address Edge, Enterprise and 5G data-centers. This allows Innovium to deliver capabilities one process node earlier than alternatives along with unique top-to-bottom scalability, which translates to faster time to market and lower R&D investment for customers.
