NVIDIA ConnectX-6 MCX653106A-HDAT 200Gb/s Dual-Port InfiniBand Smart Adapter

Engineered for demanding HPC, AI, and hyperscale cloud infrastructures, the NVIDIA® ConnectX®-6 MCX653106A-HDAT smart adapter card delivers up to 200Gb/s bandwidth per port with In-Network Computing acceleration. Offloading computation from the CPU, it dramatically improves efficiency, scalability, and security — from deep neural network training to real-time data analytics.

As a core component of the NVIDIA Quantum InfiniBand platform, ConnectX-6 enables end-to-end RDMA, hardware-based reliable transport, and advanced congestion control. The MCX653106A-HDAT model features dual-port QSFP56, supporting both InfiniBand and Ethernet (up to 200Gb/s). It integrates block-level XTS-AES encryption, NVMe over Fabrics (NVMe-oF) offloads, and GPUDirect RDMA acceleration — making it the ideal choice for GPU-accelerated clusters, software-defined storage, and virtualized networks.

NVIDIA ConnectX-6 extends Remote Direct Memory Access (RDMA) beyond conventional limits. By implementing hardware offloads for MPI tag matching, out-of-order RDMA supporting Adaptive Routing, and Dynamically Connected Transport (DCT), it ensures efficient scaling across thousands of nodes. The adapter’s In-Network Memory capability enables registration-free RDMA memory access, reducing software overhead. Combined with PCIe Gen 4.0, data moves directly between memory and network, freeing CPU cycles for application logic.

With support for RoCE (RDMA over Converged Ethernet) and overlay network tunneling offloads, ConnectX-6 provides a unified smart fabric for both InfiniBand and Ethernet environments.

• High Performance Computing (HPC): Large-scale simulations, weather modeling, and research clusters requiring low latency and high message rate.
• AI & Machine Learning: Accelerate distributed training of deep neural networks with GPUDirect RDMA and high-throughput 200Gb/s links.
• NVMe-oF Storage Arrays: Build high-performance NVMe/TCP or NVMe/RDMA storage targets with hardware offloads, reducing CPU load.
• Hyperscale Cloud & NFV: Efficient service chaining, OVS offload (ASAP²), and SR-IOV for up to 1K virtual functions per adapter.
• Big Data Analytics: In-network computing acceleration for streaming engines and distributed databases.

ConnectX-6 MCX653106A-HDAT is compatible with a wide range of servers, switches, and OS environments. It supports InfiniBand switches up to 200Gb/s (HDR) and Ethernet switches up to 200Gb/s with auto-negotiation. The adapter works across x86, Power, Arm, GPU, and FPGA-based platforms.

Note: For variants with cold plate for liquid-cooled Intel Server System D50TNP, please contact Starsurge for customized ordering.

Hong Kong Starsurge Group provides end-to-end support: from compatibility checks, firmware customization, to on-site deployment guidance. We offer dedicated technical account managers for data center upgrades and proof-of-concept (PoC) testing. All adapters are shipped with anti-static packaging and optional installation kits.
✔ 24h engineering support ticketing system ✔ Advanced replacement for business-critical environments ✔ Driver & software stack assistance (OFED, WinOF-2, DPDK).

• Confirm server mechanical clearance: standard height PCIe bracket included; low-profile bracket also provided as accessory.
• For liquid-cooled platforms (Intel D50TNP), verify cold plate option availability before ordering.
• Please confirm driver compatibility with your Linux distribution version — NVIDIA OFED recommended.
• Not publicly specified: Exact power consumption per port at full 200Gb/s load — refer to NVIDIA user manual or contact Starsurge for typical values (approx 15-18W total).
• FIPS certification is hardware-capable but may require specific firmware — notify sales team if FIPS compliance is mandatory.

• ☑ PCIe slot type: x16 mechanical (electrical x16/x8/x4 supported)
• ☑ Required port speed: 200Gb/s or lower; cable type (QSFP56 passive/active)
• ☑ OS driver availability: Check MLNX_OFED or WinOF-2 version
• ☑ Encryption requirement: FIPS mode or standard AES-XTS
• ☑ Cooling and bracket: standard or cold plate option needed?
• ☑ Quantity and lead time: Stock confirmation with Starsurge sales

• ▸ NVIDIA ConnectX-6 User Manual (Firmware & Configuration)
• ▸ RDMA over Converged Ethernet (RoCE) Deployment Guide
• ▸ NVMe-oF with ConnectX-6: Best Practices
• ▸ Performance Tuning for MPI and GPUDirect
• ▸ Block-level Encryption Setup for FIPS environments

* Specifications and features are based on published NVIDIA datasheet and may be updated. For exact technical details, please refer to official NVIDIA documentation or contact Starsurge pre-sales engineering.