
Add to Cart
The NVIDIA Quantum-2 QM9790 Series represents a paradigm shift in high-performance data center networking, delivering unprecedented bandwidth and computational capabilities for next-generation AI and HPC environments. This cutting-edge InfiniBand switch platform offers 64 ports of NDR 400Gb/s connectivity in a compact 1U form factor, providing exceptional port density and performance efficiency. With a total bidirectional bandwidth of 51.2 Tb/s and advanced in-network computing capabilities, the QM9790 enables researchers and data scientists to tackle increasingly complex computational challenges with reduced latency and improved efficiency.
Specification Category | Details |
---|---|
Port Configuration | 64×400Gb/s NDR InfiniBand ports (32 OSFP connectors) |
Switching Capacity | 51.2 Tb/s aggregate bidirectional bandwidth |
Form Factor | 1U rack mount (43.6mm height) |
Power Consumption | 1084W (with passive cables) / 1720W (with active optical cables) |
Management System | MLNX-OS with CLI, WebUI, SNMP, and JSON interfaces |
Physical Dimensions | 1.7" H × 17" W × 26" D (43.6×438×660.4 mm) |
Weight | 14.5 kg (31.97 lbs) |
Third-generation Scalable Hierarchical Aggregation and Reduction Protocol delivers 32× AI acceleration improvement over previous generations, supporting up to 64 parallel streams for massive data aggregation directly within the network infrastructure.
Advanced features including adaptive routing, network self-healing, quality of service (QoS), enhanced virtual lane mapping, and sophisticated congestion control mechanisms ensure optimal application performance under varying load conditions.
Accelerate deep learning and machine learning workloads with in-network computing capabilities
Support scientific simulations and complex computational research with ultra-low latency
Enable high-performance cloud services and big data analytics platforms
Support high-frequency trading and real-time risk analysis applications
Model Number | Port Configuration | Airflow Direction | Management |
---|---|---|---|
MQM9790-NS2F | 64×400Gb/s | Front to Rear | Built-in |
MQM9790-NS2R | 64×400Gb/s | Rear to Front | Built-in |
MQM9790-NS2FS | 64×400Gb/s | Front to Rear | Built-in (Secure Boot) |
MQM9790-NS2RS | 64×400Gb/s | Rear to Front | Built-in (Secure Boot) |
1-year comprehensive manufacturer warranty with optional extension plans available
24/7 technical support with dedicated engineering resources for critical issues
SHARPv3 (Scalable Hierarchical Aggregation and Reduction Protocol) represents NVIDIA's third-generation in-network computing technology, delivering 32× improvement in AI acceleration by enabling data aggregation within the network infrastructure itself, significantly reducing data movement and computational overhead.
Yes, the NVIDIA Quantum-2 platform maintains full backward compatibility with previous InfiniBand generations, including EDR and HDR, allowing seamless integration with existing infrastructure and mixed deployment environments.
The QM9790 series offers both built-in management through MLNX-OS with multiple interface options (CLI, WebUI, SNMP, JSON) and support for external management via NVIDIA UFM for advanced orchestration, monitoring, and analytics capabilities.
With over a decade of industry expertise, we operate extensive manufacturing facilities and maintain a robust technical team with deep knowledge of high-performance networking solutions. As an authorized partner for leading networking brands including Mellanox, Ruckus, Aruba, and Extreme, we provide comprehensive networking solutions including enterprise-grade network switches, high-performance network interface cards, wireless access points, and connectivity solutions.
Our substantial inventory investment (exceeding $10 million) ensures reliable supply and competitive pricing, while our 24/7 technical support team provides expert assistance for deployment, configuration, and troubleshooting. We have established a strong reputation for quality, reliability, and customer service in global markets, serving diverse industries from research institutions to enterprise data centers.