End-to-end 800 Gb/s InfiniBand platform for the largest AI training clusters.
NVIDIA Quantum-X800 is the first end-to-end 800 Gb/s networking platform, purpose-built for trillion-parameter AI. It combines Quantum-X800 InfiniBand switches, ConnectX SuperNICs and LinkX cables and transceivers.
Hardware in-network computing with SHARP v4, adaptive routing and telemetry-based congestion control keep large training jobs running at full speed across thousands of GPUs.
Double the bandwidth of the previous InfiniBand generation.
Collective operations are reduced inside the switch, freeing GPU time.
Telemetry-based congestion control keeps traffic flowing across the fabric.
Quantum-X800 switches offer 144 ports of 800 Gb/s with UFM management.
ConnectX SuperNICs with accelerated MPI engines and ultra-low latency.
Co-packaged optics reduce latency and power by shortening the optical path.
| Ports | 144 |
|---|---|
| Per-port bandwidth | 800 Gb/s |
| In-network computing | SHARP v4 |
| Features | Adaptive routing, telemetry-based congestion control, performance isolation, low-power link state |
| Management | NVIDIA UFM |
| ConnectX SuperNICs | Up to 1.6 Tb/s per GPU, accelerated MPI engines, QoS, adaptive routing |
|---|---|
| LinkX | Cables and transceivers, including passive fiber and linear active copper |
Ethernet platform purpose-built for AI, from single-cluster fabrics to multi-site AI factories with Spectrum-XGS.
SuperNIC delivering up to 800 Gb/s for InfiniBand or Ethernet GPU-to-GPU networking.
Data processing unit that offloads networking, storage and security from host CPUs using NVIDIA DOCA.
Tell us your workload and timeline for the NVIDIA Quantum-X800 InfiniBand. We reply within one business day with a configuration, pricing and lead time.