1G Ethernet IP Suite

Tri-speed Ethernet MAC, SGMII PCS, IEEE 1588 timestamping and a hardware UDP/IP stack. One vendor-independent IP family for Altera, Lattice, Microchip and AMD Xilinx FPGAs.

Overview

The RunX 1G Ethernet IP Suite connects an FPGA to a standard Gigabit Ethernet PHY and carries application data over UDP without a processor. Four cores; media interface, MAC, IPv4 stack and UDP stack can be used together as a complete network endpoint or individually. The same RTL targets every supported FPGA family; only a thin, documented I/O layer is vendor-specific. The suite is silicon-validated on both RGMII and SGMII at 10, 100 and 1000 Mb/s.

Designed for low latency, the core minimizes delay on both the ingress and egress paths, so frames are received and forwarded with almost no added overhead.

Features

  • 10/100/1000 Mbit/s : RGMII and SGMII, full-duplex, from one 125 MHz clock
  • MEDIA / MAC / IP / UDP IPs in one suite, all in PL, no CPU in the data path.
  • IEEE 1588 RX/TX hardware timestamps at all three speeds, drift-free
  • IEEE 802.3x PAUSE receive and transmit
  • ARP, ICMP, IPv4 and UDP supported
  • Designed for low-latency

Architecture

Architecture Architecture

Each core has its own AXI4-Lite register block, interrupt controller and statistics counters. Cores are usable together or standalone.

Why RunX

  • One package for MAC, RGMII, SGMII PCS, IEEE 1588 and UDP/IPv4. No third-party cores to integrate.
  • Vendor-independent RTL with a single verified code base across four FPGA vendors.
  • Own 8b/10b PCS with Clause 37 / SGMII auto-negotiation: SGMII without a vendor PCS licence.
  • Automatic SGMII speed from auto-negotiation, no software involvement.

Built to be trusted

  • Silicon-validated: All board tests passed on both RGMII and SGMII at 10/100/1000 Mbps.
  • IEEE 1588 without drift: zero timestamp slope over long turnaround runs at 10, 100 and 1000 Mb/s.
  • End-to-end integrity: Ethernet FCS, IPv4 header and UDP checksums verified in hardware; bad frames are dropped and counted.
  • Lossless under load: IEEE 802.3x flow control proven with a zero-loss soak test on hardware.

4 IPs, One Suite

MEDIA Interface

  • RGMII, 10/100/1000, single 125 MHz clock with clock enable
  • SGMII with in-house 8b/10b PCS and Clause 37 / SGMII auto-negotiation; automatic speed selection
  • MDIO Clause 22 master for PHY management
  • Thin vendor I/O layer: Altera, Lattice, Microchip and AMD Xilinx DDR cells

Ethernet MAC

  • Full-duplex 10/100/1000; FCS (CRC-32) generation and checking, padding, programmable IFG, separate RX/TX maximum frame size.
  • IEEE 802.3x PAUSE receive and transmit, FIFO-threshold driven
  • IEEE 1588 time-of-day counter with RX/TX hardware timestamps
  • Unicast / broadcast / promiscuous filtering; 17 statistics counters

IPv4 Stack

  • ARP reply, request and cache (1 to 8 entries); ICMP echo reply
  • IPv4 receive/transmit with header checksum; 26 statistics counters: traffic, ARP, ICMP and 9 per-cause drop reasons
  • Store-and-forward frame FIFO (2 to 8 KB) with flow control

UDP Stack

  • Four port channels plus catch-all
  • UDP checksum generation and checking
  • AXI4-Stream data, AXI4-Lite control, interrupts, 15 statistics counters

Supported FPGA Vendors

Our RTL is vendor agnostic. Altera, Lattice, Microchip, AMD Xilinx FPGA vendors are supported and reference designs are ready for integrating your project.

Resource Utilization

Each cell shows the minimum / full-featured configuration.

Core (minimum / full-featured)LUTsRegistersBlock RAM
Media interface (GMII / SGMII + tri-speed)150 / 704208 / 5710 / 0
Ethernet MAC (base / PTP + PAUSE + tri-speed)1,602 / 2,4791,417 / 2,2920 / 0
IPv4 stack (no FIFO, ARP 1 / 8 KB FIFO, ARP 8)2,136 / 2,5702,273 / 3,0320 / 2.5
UDP stack (base / with checksum)1,174 / 1,479989 / 1,1840 / 0.5

AMD Artix-7 xc7a100t-2, Vivado 2025.2, core logic only, 125 MHz. Block RAM in 36 Kb tiles.

For Evaluation, Pricing and Licensing; Contact us!

info@run-x.com