NVIDIA Mellanox MCX345A-FCPN ConnectX-3 Pro EN Single-Port QSFP+ 40GbE OCP 2.0 Mezzanine Card: This is a refurbished OCP 2.0 mezzanine network card with a single QSFP+ port at 40 Gigabit Ethernet, built on the ConnectX-3 Pro EN controller. The host interface is PCIe 3.0 x8 routed through the OCP mezzanine connector. This is NOT a standard PCIe add-in card and will not fit a normal PCIe slot. The EN suffix means Ethernet-only — InfiniBand and VPI mode are not supported on this SKU. This is a refurbished unit.
Engineering Context
OCP (Open Compute Project) Mezzanine 2.0 is the network form factor used by hyperscale and OEM servers to centralize NICs and free up rear PCIe slots for storage. ConnectX-3 Pro adds VXLAN and NVGRE stateless overlay-network offload on top of the ConnectX-3 baseline, making it relevant for SDN-managed virtualization. PCIe 3.0 x8 provides 64Gb/s of host bandwidth, enough to run a single 40GbE port at line rate. The QSFP+ port supports 40GbE (4x10G NRZ); ConnectX-3 Pro does not support 56GbE on this Pro EN SKU. Power draw is approximately 9W typical.
Deployment & Use Cases
OCP-server fleet refresh where 40GbE uplinks per node are required. Storage front-end on OCP servers using NFS, iSCSI, or RoCEv2 NVMe-oF at 40GbE. Hyper-converged infrastructure nodes needing overlay-network offload (VXLAN, NVGRE) at the host NIC. Compute nodes on dense 40GbE leaf-spine fabrics where each server gets a single uplink.
Technical Specifications
OEM: NVIDIA Mellanox
Part Number: MCX345A-FCPN
Controller: ConnectX-3 Pro EN
Ports: Single QSFP+
Data Rate: 40 Gigabit Ethernet (40GbE)
Host Interface: PCIe 3.0 x8 via OCP connector
Form Factor: OCP 2.0 Mezzanine
RDMA: RoCEv1, RoCEv2 with congestion control
Overlay Offload: VXLAN, NVGRE stateless encapsulation
Virtualization: SR-IOV up to 127 VFs
Protocol: Ethernet only (no InfiniBand)
Condition: Refurbished — Tested by T.E.S IT-SOLUTIONS
Compatibility & Constraints
Requires OCP 2.0 mezzanine slot in the host server. WILL NOT fit standard PCIe x8/x16 add-in slots. Not mechanically compatible with OCP 3.0 (NIC 3.0) cages. Server BIOS must enumerate the OCP slot for the card to be discovered. This is the EN variant — InfiniBand and VPI mode are not selectable in firmware. QSFP+ port is 40GbE only — does not support 100GbE or 56GbE. QSFP+ cage accepts passive DAC, active DAC, AOC, and SR4/LR4 optical transceivers; vendor-coded transceiver enforcement may apply depending on OEM server BIOS. RoCEv2 in production requires PFC/DCB on the switching fabric.
Why TES
T.E.S IT-SOLUTIONS bench-tests every OCP 40GbE NIC for QSFP+ link establishment, PCIe x8 width through the OCP connector, and firmware reachability before shipment. Visit tes-itsolutions.com for compatibility checks against your specific OCP server platform. Handling time 1–3 business days with international shipping.