NVIDIA ConnectX Smart Adapter for AI, HPC and Data Centers

• AI and GPU cluster networking
• High-performance computing
• Enterprise data-center connectivity
• Cloud and hyperscale infrastructure
• Accelerated storage networking
• NVMe over Fabrics deployments
• Clustered database environments
• Machine learning infrastructure
• Scientific computing platforms

• High-performance server networking
• InfiniBand connectivity across supported models
• Ethernet connectivity across supported models
• Low-latency communication
• In-Network Computing capabilities on supported generations
• AI and HPC workload optimization
• Cloud and hyperscale deployment support
• Accelerated storage connectivity
• Scalable network architecture
• High-bandwidth GPU cluster connectivity
• Enterprise data-center integration
• Network efficiency optimization
• Supports multiple ConnectX generations
• Available in different port and speed configurations 

• Brand: NVIDIA
• Product Family: ConnectX
• Product Type: Smart Network Adapter
• Network Technologies: InfiniBand and Ethernet
• Primary Applications: AI, HPC, Cloud, Storage, Data Centers
• Maximum Family Throughput: Up to 1.6Tb/s on ConnectX-9
• ConnectX-8 Throughput: Up to 800Gb/s
• ConnectX-7 Throughput: Up to 400Gb/s
• Adapter Type: HCA / Network Adapter
• Connectivity: Server and data-center networking 

WHY CHOOSE THIS PRODUCT?

• NVIDIA ConnectX is suited to organizations upgrading server, AI, HPC or storage networking where high bandwidth and low latency are priorities. Its main procurement advantage is the broad family range, allowing buyers to select the appropriate generation, speed and protocol instead of adopting a one-size-fits-all adapter. Exact compatibility should be validated against the required ConnectX model, server PCIe interface, operating system, network technology and switch infrastructure before purchase.