NVIDIA ConnectX-5 InfiniBand_Ethernet Socket Direct Adapter Cards User Manual

This is the User Guide for InfiniBand/Ethernet adapter cards based on the ConnectX®-5 integrated circuit device. NVIDIA offers an alternate ConnectX-5 Socket Direct™ card for servers without x16 PCIe slots. The adapter’s 16-lane PCIe bus is split into two 8-lane buses, with one bus accessible through a PCIe x8 edge connector and the other bus through an x8 parallel connector to an Auxiliary PCIe Connection Card. The two cards should be installed into two adjacent PCIe x8 slots and connected using a dedicated harness.


Since the two PCIe cards are installed in two PCIe slots, each card gets a unique PCI Bus and Device number. Each of the PCIe x8 busses sees two network ports; in effect, the two 100Gb/s physical ports of the ConnectX-5 Socket Direct Adapter are viewed as four netdevices by the system, with each netdevice providing a maximum of 50Gb/s data transfer rate. In other words, 100Gb/s on a (physical) adapter port (Port 1 or Port 2) will be split on the PCIe side to 50Gb/s on each of the two PCIe cards (the Adapter Card and the Auxiliary PCIe Connection Cards, respectively, in the above picture).

ConnectX-5 InfiniBand/Ethernet Socket Direct Adapter Cards


ConnectX-5 InfiniBand/Ethernet Adapter Cards

Part Number




Low Profile Adapter Card Size: 6.6 in. x 2.71 in. (167.65mm x 68.90mm)
Auxiliary PCIe Connection Card Size: 4.44 in. x 1.57 in. (113.0mm x 40.00mm)

Slim-Line SAS Cable: 9.8 in. x 1.02 in. (250cm x 25.95mm)

Slim-Line SAS Cable: 9.8 in. x 1.02 in. (350cm x 25.95mm)

Network Connector Type


Data Rate

Ethernet: 10/25/40/50/100 Gb/s

PCI Express Connectors

PCIe Gen 3.0 x8; SerDes @ 8.0GT/s


RoHS Compliant

Adapter IC Part Number


Device ID (decimal)

4119 for Physical Function (PF) and 4120 for Virtual Function (VF)


Full performance of the adapter card (100Gb/s) is possible only when installing both the adapter card and the auxiliary PCIe connection card in the dual PCIe x8 connectors.
Please refer to the performance optimization document for further guidance:

For more detailed information see Specifications.


This section describes hardware features and capabilities. Please refer to the relevant driver and/or firmware release notes for feature availability.



PCI Express (PCIe)

Uses PCIe Gen 3.0 (8GT/s) through two x8 edge connectors. Gen 1.1 and 2.0 compatible.

100Gb/s InfiniBand/Ethernet Adapter

ConnectX-5 offers the highest throughput Ethernet adapter, supporting EDR 100Gb/s InfiniBand and 100Gb/s Ethernet and enabling any standard networking, clustering, or storage to operate seamlessly over any converged network leveraging a consolidated software stack.

InfiniBand Architecture Specification v1.3 compliant

ConnectX-5 delivers low latency, high bandwidth, and computing efficiency for performance-driven server and storage clustering applications. ConnectX-5 is InfiniBand Architecture Specification v1.3 compliant.

EDR InfiniBand

A standard InfiniBand data rate, where each lane of a 4X port runs a bit rate of 25.78125Gb/s with a 64b/66b encoding, resulting in an effective bandwidth of 100Gb/s.

Up to 100 Gigabit Ethernet

NVIDIA adapters comply with the following IEEE 802.3 standards:

  • 100GbE/ 50GbE / 40GbE / 25GbE / 10GbE / 1GbE

  • IEEE 802.3bj, 802.3bm 100 Gigabit Ethernet

  • IEEE 802.3by, Ethernet Consortium25, 50 Gigabit Ethernet, supporting all FEC modes

  • IEEE 802.3ba 40 Gigabit Ethernet

  • IEEE 802.3ae 10 Gigabit Ethernet

  • IEEE 802.3ap based auto-negotiation and KR startup

  • Proprietary Ethernet protocols (20/40GBASE-R2, 50GBASE-R4)

  • IEEE 802.3ad, 802.1AX Link Aggregation

  • IEEE 802.1Q, 802.1P VLAN tags and priority

  • IEEE 802.1Qau (QCN)

  • Congestion Notification

  • IEEE 802.1Qaz (ETS)

  • IEEE 802.1Qbb (PFC)

  • IEEE 802.1Qbg

  • IEEE 1588v2

  • Jumbo frame support (9.6KB)


  • PCI Express - stores and accesses InfiniBand and/or Ethernet fabric connection information and packet data.

  • SPI Quad - includes 128Mbit SPI Quad Flash device (W25Q128FVSIG device by ST Microelectronics)

  • FRU EEPROM - Stores the parameters and personality of the card. The EEPROM capacity is 128Kbit. FRU I2C address is (0x50) and is accessible through the PCIe SMBus. (Note: A ddress 0x58 is reserved.)

Overlay Networks

In order to better scale their networks, data center operators often create overlay networks that carry traffic from individual virtual machines over logical tunnels in encapsulated formats such as NVGRE and VXLAN. While this solves network scalability issues, it hides the TCP packet from the hardware offloading engines, placing higher loads on the host CPU. ConnectX-5 effectively addresses this by providing advanced NVGRE and VXLAN hardware offloading engines that encapsulate and de-capsulate the overlay protocol.

RDMA and RDMA over Converged Ethernet (RoCE)

ConnectX-5, utilizing IBTA RDMA (Remote Data Memory Access) and RoCE (RDMA over Converged Ethernet) technology, delivers low-latency and high performance over Band and Ethernet networks. Leveraging data center bridging (DCB) capabilities as well as ConnectX-5 advanced congestion control hardware mechanisms, RoCE provides efficient low-latency RDMA services over Layer 2 and Layer 3 networks.

NVIDIA PeerDirect™

PeerDirect™ communication provides high efficiency RDMA access by eliminating unnecessary internal data copies between components on the PCIe bus (for example, from GPU to CPU), and therefore significantly reduces application run time. ConnectX-5 advanced acceleration technology enables higher cluster efficiency and scalability to tens of thousands of nodes.

CPU Offload

Adapter functionality enabling reduced CPU overhead allowing more available CPU for computation tasks.

Open VSwitch (OVS) offload using ASAP2

• Flexible match-action flow tables
• Tunneling encapsulation / decapsulation

Quality of Service (QoS)

Support for port-based Quality of Service enabling various application requirements for latency and SLA.

Hardware-based I/O Virtualization

ConnectX-5 provides dedicated adapter resources and guaranteed isolation and protection for virtual machines within the server.

Storage Acceleration

A consolidated compute and storage network achieves significant cost-performance advantages over multi-fabric networks. Standard block and file access protocols can leverage InfiniBand RDMA for high-performance storage access.

  • NVMe over Fabric offloads for target machine

  • Erasure Coding

  • T10-DIF Signature Handover


ConnectX-5 SR-IOV technology provides dedicated adapter resources and guaranteed isolation and protection for virtual machines (VM) within the server.

High-Performance Accelerations

  • Tag Matching and Rendezvous Offloads

  • Adaptive Routing on Reliable Transport

  • Burst Buffer Offloads for Background Checkpointing

Supported in OpenFabrics Enterprise Distribution (OFED).

  • Interoperable with 1/10/25/40/50/100 Gb/s Ethernet switches

  • Passive copper cable with ESD protection

  • Powered connectors for optical and active cable support

Socket Direct technology maintains support for manageability through a BMC. The Socket Direct PCIe stand-up adapter can be connected to a BMC using MCTP over SMBus or MCTP over PCIe protocols as if it is a standard NVIDIA PCIe stand-up adapter. For configuring the adapter for the specific manageability solution in use by the server, please contact NVIDIA Support.

© Copyright 2023, NVIDIA. Last updated on May 22, 2023.