NV-CONNECTX-SNIC-1600

NVIDIA ConnectX NICs and SuperNICs

High-performance intelligent network adapters delivering ultra-low latency, RDMA acceleration, and up to 1.6 Tb/s throughput for AI, HPC, and cloud data centers.

System Metrics Table

Architecture NVIDIA ConnectX
Connectivity Ethernet and InfiniBand
Deployment AI Factories, Enterprise Data Centers, Cloud Platforms, HPC Clusters
Form Factor PCIe Half-Height Half-Length and OCP 3.0
GPUDirect GPUDirect Storage and GPUDirect RDMA
In-Network Computing Hardware-Accelerated AI Communication
Latency Ultra-Low Latency
Management Integrated with NVIDIA Networking Software Ecosystem
Maximum Throughput Up to 1.6 Tb/s
Network Offloads TCP/IP, RoCE, NVMe-oF, Storage and Security Acceleration
PCI Express Interface PCIe Gen6 x16 (Backward Compatible with Gen5)
Platform Type Network Interface Card (NIC) / SuperNIC
Port Configuration Single-Port or Dual-Port High-Speed Interfaces
Power Efficiency Hardware Offloading Reduces CPU Utilization and Energy Consumption
Primary Workloads Generative AI, Machine Learning, HPC, Storage Networking, Cloud Computing
Protocol Support RoCE, InfiniBand, TCP/IP, UDP, NVMe over Fabrics
RDMA Support Native RDMA and GPUDirect RDMA
Scalability Designed for Hyperscale GPU Clusters
Security Secure Boot and Hardware Cryptographic Acceleration