Products 2018-05-22T05:15:22+00:00

Aries MT3250 Family

  • Greater architectural scalability, ranging from 480G to 960G and >900K IP addresses with >80K ACL all under one efficient and familiar software eco system
  • Hyper-efficient solutions for next generation datacenters: designed from the start to provide joules/gigabit over all IC fabrication nodes
  • Advanced purpose-built deterministic pipeline addresses the extremely cost sensitive access 2.5G/10G Enterprise and Carrier client connectivity markets
  • Micro-code programmable 28nm switch to reach mass production (first ASIC A0 revision went MP CY Q2 2016)
  • Programmable deep packet classification, programmable recursive match, result and actions, efficient granular and fully fungible match/result table memories, programmable packet generation/modification including real-time visibility instruments and fast path based advanced diagnostics
  • Full product line with unique price (many unique cost points) allow perfect fit of power, price and performance for all data center tiers and applications
  • Lower TCO achieved through simplicity and leading architectural efficiency: A perfect fit of power-price-performance matched to any Datacenter size/shape
  • Lower TCO possible for Ethernet networking infrastructure: Lower CAPEX: less PCB layers, smaller power supply, smaller software image for minimum DRAM at the host CPU. Lower OPEX: Maximum uptime due to advanced real-time programmable troubleshooting, analytics, instrumentation & packet based telemetry yet maintains the lower energy bill (lower typical, average and peak power per gigabit).

Aries Hybrid ToR Switch

  • 480-960 Gbps duplex I/O bandwidth
  • Wire-speed 1.44 BPPS at 64B frame size
  • Up to 96 High speed 10GHz SerDes
  • Low Latency Cut-through and Store Forward
  • Hybrid Switch Mode Flow Based SDN Architecture
  • L2/L3 Forwarding and Tunneling, advanced underlays and overlays
  • Flexible Port Configuration: e.g. 48 x 10G + 4 x 100G, 24 x 40G, 8 x 100G, 96 x10G
  • 28nm HPM TSMC

Taurus Family NP8360 Series

  • Advanced purpose-built architecture designed to addresses next generation network automation (real-time visibility/switch table state modification) for hyper-scale markets 1.08T to 6.4T (10G/25G/40G/50G/100G NRZ connectivity)
  • Pioneering the disaggregated network componentry era: Production with ‘Network scale multi-die package technology’ allowing higher yields, better price-performance fit for the absolute lower TCO
  • Taurus 6.4T multi-die package debut, greater efficiency requires we all, “do more with less”
  • Full wider product line range, offering more unique price points (many different individual unique power, scale and cost points) thus allow perfect fit of power, price and performance for all data center tiers and applications
  • Wire-speed programmable deep packet classification, flexible and recursive match/result-actions, packet generation/modification, fungible match and result memory including real-time instrumentation, packet diagnostics and state updates
  • Advanced L2-L5 programmable entropy with micro and jumbo flow-steering for better fabric utilization
  • Larger Service Scale: >800K IP addresses, >70K ACLs, push > 10 labels (MPLS or user defined)
  • Granular Fungible Memories for all Tables and database result Memories (Flexible Shared Data structures for Forwarding, QoS, Instrumentation and Security)
  • Lower power of any mass production released 16nm 6.4T NRZ merchant ASIC devices in the market

Only one familiar software and one engineering porting expenditure will yield a complete comprehensive system portfolio, spanning 480G – 6.4T at lower TCO

Cloud gets its power from scale. 

Scale requires great efficiency. 

Taurus is both physically and philosophically perfectly aligned for achieving greater cloud scale at the lower TCO.

Note : “Absolute maximum power dissipation can only be achieve in a Nephos (USA) Inc. Laboratory under specific non-real world traffic conditions”

Taurus ToR and Fabrics Switch

  • Product derivatives with wide range of bandwidth spanning: 880G to 6.4T
  • Up to 256 high speed 28GHz NRZ SerDes
  • Large intelligent packet buffer for superior in-cast performance
  • Buffer size options: 20, 28, 40, and >50 MB (virtual queue pool, programmable dynamic allocation, intelligent hysteretic water marks)
  • Flexible Port Configuration: 10/25/40/50/100G NRZ
  • 2 x 10G. 2x 1G and 2 x 2.5G dedicated management ports, PCIe Gen3 (>30G)
  • 16 nm FinFET+Multi-die packaging enhanced patented silicon interposer TX/RX
  • Available Now: Q2 2017, GA Sept 2017

SDK Features

  • Extensibility
  • Hardware Abstraction Layer across multiple chips
  • Portability
  • OS-independent: OS Abstraction Layer
  • Endian agnostic
  • 64-bit-addressing
  • Easy to Debug
  • APIs to trouble-shoot SDK problems
  • APIs for chip diagnostic

NPOS Architecture

Looking for more detailed product or commercial information?

Please email us at :