 
|
NVIDIA Mellanox ConnectX-8 C8180 900-9X81E-00EX-ST0 800G InfiniBand/Ethernet Adapter
Product Details:
| Brand Name: | Mellanox |
| Model Number: | C8180(900-9X81E-00EX-ST0) |
| Document: | connectx-datasheet-connectx...1).pdf |
Payment & Shipping Terms:
| Minimum Order Quantity: | 1pcs |
|---|---|
| Price: | Negotiate |
| Packaging Details: | outer box |
| Delivery Time: | Based on inventory |
| Payment Terms: | T/T |
| Supply Ability: | Supply by project/batch |
|
Detail Information |
|||
| Model NO.: | C8180(900-9X81E-00EX-ST0) | Transmission Rate: | 400gbe |
|---|---|---|---|
| Ports: | Dual-Port | Function: | Lacp, Stackable, Vlan Support |
| Technology: | Infiniband | Host Interface: | Gen6 X16 |
| Interface Type: | Qsfp112 | Trademark: | Mellanox |
| Highlight: | NVIDIA Mellanox ConnectX-8 network adapter,800G InfiniBand Ethernet adapter,Nvidia network card with warranty |
||
Product Description
NVIDIA ConnectX-8 SuperNIC C8180
800Gb/s AI Fabric Adapter for Hyperscale Data Centers
Hero Summary
The NVIDIA ConnectX-8 SuperNIC C8180 (900-9X81E-00EX-ST0) is a single-port OSFP adapter engineered for the most demanding AI and high-performance computing environments. Delivering up to 800Gb/s bandwidth over InfiniBand or Ethernet, it eliminates network bottlenecks in large-scale GPU clusters. With a PCIe 6.0 x16 host interface, advanced RDMA acceleration (GPUDirect), and hardware-based security, the C8180 is the cornerstone of next-generation AI factories and cloud data centers.
Product Overview
The NVIDIA ConnectX-8 C8180 SuperNIC (900-9X81E-00EX-ST0) represents a generational leap in network acceleration. Optimized for hyperscale AI workloads, it provides up to 800 Gb/s bi-directional bandwidth, a native PCIe Gen6 host interface (up to 48 lanes), and advanced telemetry-based congestion control. Built for both InfiniBand and Ethernet fabrics, this adapter enables trillion-parameter GPU computing and agentic AI applications with unprecedented efficiency.
The ST0 (Standard Enterprise) variant is pre-validated for out-of-box compatibility with leading server OEMs, emphasizing deployment simplicity, global regulatory compliance, and extended lifecycle support. It is ideal for organizations migrating from ConnectX-6/7 to 800G fabrics with minimal integration friction.
Key Facts
- Brand: NVIDIA / Mellanox
- Model: C8180 (900-9X81E-00EX-ST0)
- Port: 1x OSFP (Split to 2x 400G or 8x 100G)
- Host Bus: PCIe 6.0 x16 (64 GT/s)
- Form Factor: HHHL (68.50mm x 168.40mm)
- Security: Secure Boot, Hardware Root of Trust, Inline Crypto (IPsec/MACsec/PSP)
Key Features
800Gb/s Bandwidth
Supports XDR InfiniBand (800/400/200/100 Gb/s) and Ethernet (400/200/100/50/25 Gb/s) for maximum deployment flexibility.
PCIe 6.0 x16 Interface
Delivers 64 GT/s link rate with support for SocketDirect/Multi-Host, enabling up to four hosts to share a single adapter.
GPUDirect RDMA & Storage
Enables direct memory access between GPU and storage, eliminating copy overhead and reducing latency for AI/ML workloads.
In-Network Computing (SHARPv3)
Offloads collective operations (MPI) to the network, accelerating HPC and AI training by reducing data movement.
Hardware Security
Integrates Secure Boot, Flash Encryption, Device Attestation (SPDM 1.1), and Inline Crypto (IPsec/MACsec/PSP) to safeguard data-in-transit.
ASAP² & Overlay Acceleration
Accelerated Switching and Packet Processing for SDN/OVS, with hardware offload for VXLAN, GENEVE, and NVGRE overlays.
Technology
- RDMA & RoCEv2: Advanced, programmable congestion control for low-latency, high-throughput fabrics.
- Ethernet Accelerated Switching and Packet Processing (ASAP²): Offloads SDN and VNF workloads, optimizing cloud networking efficiency.
- Overlay Network Acceleration: Hardware support for VXLAN, GENEVE, and NVGRE with hierarchical QoS and flow mirroring.
- Stateless TCP Offloads: Includes LSO, LRO, GRO, TSS, and RSS to reduce CPU overhead.
- Precision Timing: IEEE 1588v2 Class C, SyncE, and PTM for time-sensitive applications.
Typical Deployments
Compatibility
| Category | Supported Options / Details |
|---|---|
| Operating Systems | RHEL, Ubuntu, SUSE, Windows Server, WinOF-2 |
| Software Stacks | NVIDIA NCCL, HPC-X, DOCA UCC/UCX, Open MPI, MVAPICH2 |
| Server Platforms | Pre-validated with HPE, Supermicro, and major OEM servers |
| Certifications | CE, FCC, UKCA, BSMI, VCCI (RoHS, REACH compliant) |
Compatibility Matrix
- InfiniBand: 1x 800/400/200/100 Gb/s
- Ethernet: 1x 400/200/100/50/25 Gb/s or 2x 400/200/100/50/25 Gb/s (port split)
- PCIe: Gen6 x16, backward compatible with Gen5/Gen4
Specification Table
| Parameter | Specification |
|---|---|
| Brand | NVIDIA (Mellanox) |
| Model | C8180 (900-9X81E-00EX-ST0) |
| Port Configuration | Single-cage OSFP |
| InfiniBand Speed | Up to 800 Gb/s XDR (default mode) |
| Ethernet Speed | 2x 400GbE (port split) |
| Host Interface | PCIe 6.0 x16 (up to 48 lanes) |
| Extension Options | SocketDirect / Multi-Host |
| Form Factor | HHHL (68.50mm x 168.40mm) |
| Security | Crypto & Secure Boot enabled |
| Operating Temperature | 0°C to 55°C (32°F to 131°F) |
| Storage Temperature | -40°C to 70°C (-40°F to 158°F) |
Specifications are for reference. Please confirm full SKU details before ordering.
Selection Guide
- Choose the C8180 ST0 variant if you require:
- Enterprise-optimized SKU with pre-validated driver stacks for RHEL, Ubuntu, and Windows Server.
- Global regulatory certifications (CE, FCC, UKCA) for seamless deployment across regions.
- Extended firmware support lifecycle (5+ years) and standard 3-year warranty.
- For high-density GPU clusters requiring maximum bandwidth and low latency, the C8180 is the premier choice. For downstream port expansion, consider other variants in the ConnectX-8 family.
Advantages
- Unmatched Bandwidth: 800Gb/s total throughput for data-intensive AI/ML workloads.
- Ultra-Low Latency: RDMA and GPUDirect acceleration for real-time analytics and training.
- Protocol Flexibility: Native support for both InfiniBand and Ethernet fabrics.
- Investment Protection: PCIe 6.0 x16 interface with backward compatibility to PCIe 5.0/4.0.
- Security-First Design: Hardware-based root of trust and inline encryption protect sensitive data.
Service & Support
- Standard 3-year warranty with optional extension.
- Global regulatory compliance (CE, FCC, UKCA, VCCI, BSMI).
- Access to NVIDIA Enterprise Marketplace for partner support and services.
- Firmware updates and technical documentation available via NVIDIA support portal.
Buyer Checklist
- Confirm server has PCIe 6.0/5.0 x16 slot with sufficient power and cooling (0-55°C ambient).
- Select OSFP optical modules or DAC/AOC compatible with 800G/400G PAM4 signaling.
- Verify operating system and driver support (RHEL, Ubuntu, Windows).
- Plan network topology for InfiniBand or Ethernet fabric, considering port split (2x 400G or 8x 100G).
- Check for RDMA and GPUDirect enablement in BIOS and software stack.
Frequently Asked Questions (FAQ)
Precautions
- PCIe Channel Allocation: Ensure the server provides a x16 slot directly connected to the CPU without bifurcation to achieve maximum bandwidth.
- Module Compatibility: Use only certified OSFP modules to avoid link negotiation failures and performance degradation.
- RDMA Configuration: Properly configure BIOS (SR-IOV, ACS) and system settings (jumbo frames, interrupt affinity) for optimal RDMA performance.
- Multi-Host Mode: Follow the official configuration guide for SocketDirect to avoid conflicts with other expansion cards.
- Thermal Management: Adhere to the specified operating temperature range (0-55°C) to ensure stable operation.
About NVIDIA Mellanox
NVIDIA Mellanox (formerly Mellanox Technologies, founded in 1999) is a global leader in end-to-end Ethernet and InfiniBand intelligent interconnect solutions for servers, storage, and hyper-converged infrastructure. Headquartered in Santa Clara, CA, and Yokneam, Israel, the company was acquired by NVIDIA in 2020 for $6.9 billion, combining its high-performance networking technology with NVIDIA's AI computing platform. With over 3,000 employees and 3,700+ patents, NVIDIA Mellanox continues to drive innovation in AI factories, cloud data centers, and high-performance computing.
Related Products
- NVIDIA ConnectX-7 Adapters
- NVIDIA Quantum InfiniBand Switches
- NVIDIA Spectrum Ethernet Switches
- NVIDIA BlueField-3 DPUs
- OSFP Optical Modules & DAC Cables
Related Guides
- Deploying 800G AI Fabrics with ConnectX-8
- InfiniBand vs. Ethernet: Choosing the Right Fabric
- Optimizing RDMA for AI Workloads
- NVIDIA DOCA Software Framework Overview
- PCIe 6.0: What Data Center Architects Need to Know