Home › Products › Ethernet Network Adapters › MCX713106AS-VEAT

NVIDIA Mellanox ConnectX-7 MCX713106AS-VEAT

PCIe 5.0 x16 Dual-Port 200GbE Adapter Card with Secure Boot

P/N: MCX713106AS-VEAT

Host Interface

PCIe 5.0 x16

Port Count

Port Speed & Connector Type

200GbE, QSFP112

RDMA / Network Protocol

RoCE v2

Secure Boot

Supported

Regulatory Certifications

Product Description

The NVIDIA Mellanox ConnectX-7 MCX713106AS-VEAT is a dual-port 200GbE Ethernet adapter card built on NVIDIA's 7th-generation ConnectX architecture, delivering high-bandwidth, low-latency networking over QSFP112 connectors backed by a PCIe 5.0 x16 host interface. This variant includes hardware Secure Boot for verified firmware integrity but does not include the Crypto (encryption offload) engine found on sibling SKUs, making it a cost-effective option for deployments that require platform security assurance without the need for hardware-accelerated data-plane encryption. It carries forward ConnectX-7's full RoCE v2 offload suite and PCIe Gen5 bandwidth advantage, suited for high-throughput AI, HPC, and hyperscale cloud networking.

Key Features

Dual-port 200GbE connectivity over QSFP112 connectors
PCIe 5.0 x16 host interface for maximum host-to-adapter bandwidth
Hardware Secure Boot for verified, tamper-resistant firmware integrity
Hardware-based RoCE v2 for low-latency, CPU-efficient remote data access
SR-IOV support for network function virtualization
Advanced congestion control and quality-of-service features
Overlay network offloads (VXLAN, NVGRE, GENEVE)
High message rate optimized for AI/ML and HPC communication patterns
Cost-effective alternative to Crypto-enabled ConnectX-7 SKUs where hardware encryption is not required

Technical Specifications

Host Interface	PCIe 5.0 x16
Port Count	2
Port Speed & Connector Type	200GbE, QSFP112
RDMA / Network Protocol	RoCE v2
Secure Boot	Supported
Crypto (Encryption Offload)	Not included on this SKU
OS Support	Linux (RHEL/Ubuntu/SUSE), Windows Server, VMware ESXi, FreeBSD (typical for ConnectX-7 family)
Management	NVIDIA MFT / NEO management tools (typical)
Operating Temperature	0°C to 55°C (typical)
Hardware Warranty	Per NVIDIA/OEM channel warranty terms
Regulatory Certifications	FCC, CE, RoHS (typical for NVIDIA ConnectX adapters)
Ordering SKU	900-9X7AH-0078-ST0

Platform Compatibility

Platform	Status

Q&A

How does this model differ from the AC variant with an embedded crypto engine (MCX713106AC-VEAT)?

Both are dual-port 200GbE, PCIe 5.0 x16 adapters; the key difference is the embedded hardware crypto engine. The AC variant includes hardware crypto offload, suited for environments requiring IPsec/TLS encryption processing at the NIC level, while this AS variant omits that engine, making it a more cost-effective choice for deployments that need platform security assurance (Secure Boot) but not hardware-accelerated data-plane encryption.

What are this card's port speed and host interface specifications? Why is PCIe 5.0 required?

This card provides dual-port 200GbE connectivity via QSFP112 connectors, one of the highest per-port speeds in the ConnectX-7 family. Paired with a PCIe 5.0 x16 host interface, it delivers maximum host-to-adapter bandwidth, ensuring the host bus does not become a bottleneck when both ports run at line rate simultaneously — well suited for the most bandwidth-intensive AI training, HPC, and hyperscale cloud networking architectures.

Does this card support overlay network offload? What virtualization scenarios is it suited for?

Yes. This card offers overlay network offload for VXLAN, NVGRE, and GENEVE, moving encapsulation/decapsulation workloads off the host CPU, suited for multi-tenant cloud network architectures. It also supports SR-IOV for network function virtualization, along with advanced congestion control and Quality of Service (QoS) features to ensure service levels for different traffic types in virtualized environments.

What application scenarios are this card's RoCE v2 and high message rate design suited for?

This card provides hardware-based RoCE v2 for low-latency, CPU-efficient remote data access, and is optimized for AI/ML and HPC communication patterns with high message rate handling, suited for the frequent, small-packet communication typical of inter-node distributed training. Combined with the high bandwidth of dual-port 200GbE and advanced congestion control, this card is particularly well suited for collective communication operations across large-scale GPU clusters.

What protection does this card's Secure Boot feature provide?

This card features Secure Boot with a hardware root-of-trust, which verifies firmware signatures at adapter boot time to ensure that only authorized, untampered firmware can run, preventing malware or unauthorized firmware modifications from compromising the adapter's security. This is a platform-level security guarantee, distinct from data-plane encryption (crypto engine) protection — deployments that don't require hardware crypto offload still benefit from this firmware integrity protection.

Connectivity

No connectivity available.

Documents & Downloads

No documents available for this product.

RAID Controllers

HBA（Host Bus Adapters）

Fibre Channel Host HBA

NVMe Adapter

Battery Backup Unit（BBU）

Ethernet Network Adapters

Wireless

Transceiver

Processor／CPU

DRAM／Memory

Server Motherboard

Embedded Motherboard

CPU／AI Accelerator

CPU Expansion Adapter

Video Capture

Stream

Convert

Camera Module

Machine Vision Frame Grabber

Cable