NVIDIA ConnectX-7 NDR 400G Single-Port OSFP InfiniBand Adapter
Single-port 400G NIC delivering full per-GPU rail bandwidth for training.
We help you choose, configure, and deliver the right system — no obligation.
Configuration at a Glance
Tailored per engagement. Full technical overview below.
Configuration Options
Core specifications for this system. Every component is configurable to your workload — request a quote for a tailored build.
Collectives, MPI tag matching, all-to-all engines
Overview
The ConnectX-7 NDR 400G is a single-port OSFP InfiniBand adapter on PCIe Gen5 x16, purpose-built to give each GPU a dedicated 400Gb/s rail in dense training servers. Nexus Compute specifies, configures, and validates each card against your server platform and firmware baseline, delivered warranty-backed through NVIDIA-authorized channels.
Who This Solution Is For
Business Benefits
Full 400G per rail
A dedicated single port keeps each GPU's NDR rail uncontended for peak all-reduce throughput.
In-network offload
Hardware collective and MPI tag-matching engines free CPU cycles and reduce communication latency.
Platform-validated
Each adapter is tested against your server BIOS, slot mapping, and firmware before warranty-backed delivery.
Typical Business Use Cases
Per-GPU rail adapters in DGX-class servers
Low-latency MPI for HPC simulation
RDMA host adapter for NDR fabrics
GPUDirect storage and RDMA
Industry Applications
Technical Overview
The ConnectX-7 single-port adapter terminates one NDR 400Gb/s OSFP link over a PCIe Gen5 x16 host interface, delivering up to 330-370 million messages per second with in-network computing engines for collectives, MPI tag matching, and all-to-all. GPUDirect RDMA moves data directly between GPUs and the fabric, bypassing host memory and CPU.
| Network Port | 1x NDR 400Gb/s InfiniBand (OSFP) |
| Host Interface | PCIe 5.0 x16 |
| Message Rate | Up to ~330-370 M msg/sec |
| Acceleration | Collectives, MPI tag matching, all-to-all engines |
| GPUDirect | GPUDirect RDMA and Storage |
| Form Factor | PCIe HHHL (OSFP riding-heatsink cage) |
| Security | Secure boot, hardware root of trust |
| Warranty | Manufacturer-backed via authorized channels |
Specifications are indicative and configured to each engagement. Request a quote for a configuration tailored to your requirements.
Warranty, Support & Fulfillment
Every system ships from an authorized channel, configured and tested, with the documentation enterprise buyers need — backed by warranty and a dedicated account team.
Enterprise Warranty
Full manufacturer warranty with optional on-site, next-business-day support and extended coverage.
Authorized Channel
Sourced through Tier-1 distribution and OEM partners — never grey market. Asset & warranty records included.
Lead Time & Deployment
48-hour quotes, then configured, burn-in tested, and delivered on a committed schedule.
Nationwide Fulfillment
Coordinated logistics, rack-and-stack, and delivery wherever your infrastructure lives.
Frequently Asked Questions
Single-port or dual-port ConnectX-7 for an 8-GPU server?
Most dense training designs pair one single-port 400G adapter per GPU on a dual-rail layout; the single-port card keeps each rail at full bandwidth rather than splitting a port across GPUs.
Does this adapter require PCIe Gen5?
To reach the full 400Gb/s line rate the card uses a PCIe Gen5 x16 host slot; we confirm your server provides the required lanes and slot topology before delivery.
Will it work with the Quantum-2 QM9700/QM9790 switches?
Yes, the ConnectX-7 NDR adapter is designed to pair with Quantum-2 NDR switches over OSFP optics or DACs; we specify matching LinkX cabling for your reach.
Hardware Assistance
Configure the NVIDIA ConnectX-7 NDR 400G Single-Port OSFP InfiniBand Adapter with Nexus Compute
Tell us your requirements and a hardware specialist will help you specify, configure, and quote the right system — typically within two business days. No obligation.