The FPGA SmartNIC for Network Acceleration

  • Intel® FPGA Programmable Acceleration Card (Intel® FPGA PAC) N3000 is a highly customizable SmartNIC platform for multi-workload networking infrastructure and application acceleration. It has the right memory mixture designed for network functions, with integrated network interface card (NIC) in a small form factor that enables high throughput, low latency, low power/bit for custom networking pipeline. 
  • The Intel FPGA PAC N3000 product family include a variant that is designed to be Network Equipment Building System (NEBS)-friendly, for use in NEBS-compliant systems. To help protect systems from FPGA-hosted security exploits, the Intel FPGA PAC N3000 features a Root-of-Trust device that enables more secure loading of authorized workloads and board updates, and enforces policies to help prevent unauthorized access to critical board interfaces and flash memory. End-to-end industry standard and open source tool support allow users to quickly adapt to evolving workloads and industry standards.
  • Intel is accelerating 5G and network functions virtualization (NFV) adoption for ecosystem partners, such as Telecom Equipment Manufacturers (TEMs), Virtual Network Function (VNF) vendors, system integrators, and telcos to bring scalable and high-performance solutions to market. 

Network Functions Virtualization (NFV) Workloads Enabled by Intel® FPGA PAC N3000 

Application

Provider FPGA Advantage Additional Resources
5GCN (vEPC)  Affirmed Networks

20 Gbps/ core

50% core saving on user plane forwarding

100GbE termination

vBNG

Benu Networks

Up to 8X higher throughput with lower latency and lower jitter

vfirewall

F5 Networks

Up to 300X capacity to absorb distributed denial-of-service (DDoS) attacks

IPSec

Arrive Technologies

Up to 2X throughput with lower latency

SRV6

HCL Technologies

Up to 300% performance improvements for small packets
OVS Intel/HCL Technologies Full performance for 2x25G at 512B and above

Turbo (4G)

Intel

Support up to 27 Gbps Uplink and 3 Gbps Downlink with up to 8 virtual machines (VMs) running in host 

Contact Intel for the design example

LDPC (5G)

Intel

Support up to 15 Gbps Uplink and 5 Gbps Downlink with up to 8 VMs running in hoste† 

ORAN 7.2 Split Fronthaul

Intel

Compression algorithms including Mu-Law and BFP

Contact Intel for the design example

Note: See Additional Resources for supporting data.  For more complete information about performance and benchmark results, visit www.intel.com/benchmarks.