MCX75310AAS-NEAT（900-9X766-003N-SQ0 ）Mellanox network card MCX75310AAS-NEAT ConnectX-7 Adapter Card 400GbE/NDR Crypto DisabledMCX75310AAS-NEAT（900-9X766-003N-SQ0 ）

Mellanox network card MCX75310AAS-NEAT ConnectX-7 Adapter Card 400GbE/NDR Crypto Disabled

MCX75310AAS-NEAT（900-9X766-003N-SQ0 ）

MCX75310AAS-NEAT ConnectX® -7 family of Remote Direct Memory Access (RDMA) network adapters supports InfiniBand and Ethernet protocols and a range of speeds up to 400Gb/s. MCX75310AAS-NEAT enables a wide range of smart, scalable, and feature-rich networking solutions that address traditional enterprise needs up to the world’s most-demanding AI, scientific computing, and hyperscale cloud data center workloads.

Accelerated Networking and Security

MCX75310AAS-NEAT provides a broad set of software-defined, hardware-accelerated networking, storage, and security capabilities which enable organizations to modernize and secure their IT infrastructures. Moreover, ConnectX-7 empowers agile and high-performance solutions from edge to core data centers to clouds, all while enhancing network security and reducing the total cost of ownership.

MCX75310AAS-NEAT Specifications

InfiniBand Supported Speeds [Gb/s]	Network Ports and Cages	Host Interface	Form Factor	Part Number
Up to 400Gb/s	1x OSFP	PCle Gen 4.0/5.0 x16 With option for extension	HHAL	MCX75510AAS-NEAT
Up to 400Gb/s	1x OSFP	PCle Gen 4.0/5.0 x16	HHAL	MCX75310AAS-NEAT
Up to 400Gb/s	1x OSFP	PCle Gen 4.0/5.0 2×8 in a row	HHAL	MCX75210AAS-NEAT
Up to 200Gb/s	1x OSFP	PCle Gen 4.0/5.0 x16 With option for extension	HHAL	MCX75510AAS-HEAT
Up to 200Gb/s	1x OSFP	PCle Gen 4.0/5.0 x16	HHAL	MCX75510AAS-HEAT
Up to 200Gb/s	1x OSFP	PCle Gen 4.0/5.0 2×8 in a row	HHAL	MCX75510AAS-HEAT
Up to 200Gb/s	1x OSFP	PCle Gen 4.0/5.0 x16 With option for extension	HHAL	MCX75510AAS-HEAT
Up to 200Gb/s	2x QSFP	PCle Gen 4.0/5.0 x16 With option for extension	HHAL	MCX755106AS-HEAT

MCX75310AAS-NEAT Key Features

InfiniBand Interface

> InfiniBand Trade Association Spec 1.5 compliant

> RDMA, send/receive semantics

> 16 million input/output (IO) channels

> 256 to 4Kbyte maximum transmission unit (MTU), 2Gbyte messages

Ethernet Interface

> Up to 4 network ports supporting NRZ, PAM4 (50G and 100G), in various configurations

> Up to 400Gb/s total bandwidth

> RDMA over Converged Ethernet (RoCE)

Enhanced InfiniBand Networking

> Hardware-based reliable transport

> Extended Reliable Connected (XRC)

> Dynamically Connected Transport (DCT)

> GPUDirect® RDMA

> GPUDirect Storage

> Adaptive routing support

> Enhanced atomic operations

> Advanced memory mapping, allowing user mode registration (UMR)

> On-demand paging (ODP), including registration-free RDMA memory access

> Enhanced congestion control

> Burst buffer offload

> Single root IO virtualization (SR-IOV)

> Optimized for HPC software libraries including:

> NVIDIA HPC-X® , UCX® , UCC, NCCL, OpenMPI, MVAPICH, MPICH, OpenSHMEM, PGAS

> Collective operations offloads

> Support for NVIDIA Scalable  Hierarchical Aggregation and Reduction Protocol (SHARP)

> Rendezvous protocol offload

> In-network on-board memory

NVIDIA ConnectX®-7 Intelligent network adapter MCX75310AAS-NEAT（900-9X766-003N-SQ0 ）
(HCA) delivers outstanding network performance to address the world’s most challenging problems. ConnectX-7 InfiniBand
The network card supports ultra-low latency, 400 Gb/s throughput and the innovative NVIDIA network computing engine for high performance Computing (HPC), artificial intelligence (AI), and hyperscale cloud data centers provide the acceleration, scalability, and power needed Feature-rich technology.
High-performance computing and artificial intelligence have turned supercomputers into primary data-processing engines for widespread commercial use, Support scientific research, scientific discovery and product development. These systems are capable of performing complex simulations and are turned on by software. Writing software for a new era of AI. The NVIDIA InfiniBand network is the engine for these platforms and delivers breakthrough Performance.
ConnectX-7’s intelligent network computing acceleration engines include Collection acceleration, MPI Tag Matching, and MPI_Alltoall The engine. These performance benefits, along with standard guarantees of compatibility, ensure compute-intensive and data-intensive applications The program’s outstanding performance and scalability help users protect their data center investments.