- NVIDIA Quantum-X800 InfiniBand for infrastructure dedicated to the highest performance artificial intelligence
- NVIDIA Spectrum-X800 Ethernet for AI-optimized networking in every data center
- NVIDIA software distributes computing across Blackwell, new switches and BlueField-3 SuperNICs to power AI, data processing, HPC and cloud workloads
GTC—NVIDIA today announced a new wave of network switches, the X800 series, designed for massive artificial intelligence.
The world’s first networking platforms capable of 800Gb/s end-to-end throughput, NVIDIA Quantum-X800 InfiniBand and NVIDIA Spectrum™-X800 Ethernet push the boundaries of network performance for computing and AI workloads. They contain software that further accelerates AI, cloud, data processing and HPC applications in all types of data centers, including those that include the recently announced product line based on NVIDIA Blackwell architecture.
“NVIDIA Networking is critical to the scalability of our AI supercomputer infrastructure,” said Gilad Shainer, senior vice president of networking at NVIDIA. “NVIDIA X800 switches are end-to-end networking platforms that enable us to achieve the generative AI across trillions of parameters that is essential for new AI infrastructures.”
Early adopters of Quantum InfiniBand and Spectrum-X Ethernet include Microsoft Azure and Oracle Cloud Infrastructure.
“AI is a powerful tool for turning data into knowledge. Behind this transformation is the evolution of data centers into high-performance AI engines with increased demands on network infrastructure,” said Nidhi Chappell, vice president of AI infrastructure at Microsoft Azure. “With new integrations of NVIDIA networking solutions, Microsoft Azure will continue to build infrastructure that pushes the boundaries of AI in the cloud.”
Coreweave is also among the early adopters.
The next standard for extreme performance
The Quantum-X800 platform sets a new standard in providing the highest performance for infrastructure dedicated to artificial intelligence. Includes NVIDIA Quantum Q3400 switch and NVIDIA ConnectX®-8 SuperNIC™, which together achieve a leading end-to-end bandwidth of 800Gb/s. This is a 5x higher bandwidth capacity and a 9x increase of 14.4 Tflops within grid computing with NVIDIA’s Scalable Hierarchical Aggregation and Reduction Protocol (SHARPv4) compared to the previous generation.
The Spectrum-X800 platform provides optimized network performance for AI cloud and enterprise infrastructure. Using Spectrum SN5600 800Gb/s switches and NVIDIA BlueField®-3 SuperNIC, the Spectrum-X800 platform provides advanced feature sets critical to multi-tenant generative AI clouds and large enterprises.
The Spectrum-X800 optimizes network performance, facilitating faster processing, analysis and execution of AI workloads, thereby accelerating the development, deployment and time to market of AI solutions. Designed specifically for multi-tenant environments, the Spectrum-X800 provides performance isolation for each tenant’s AI workloads to maintain optimal and consistent performance levels, increasing user satisfaction and quality of service.
NVIDIA software support
NVIDIA offers a comprehensive suite of network acceleration libraries, software development packages, and management software to optimize performance for a trillion AI model parameters.
These include the NVIDIA Collective Communications Library (NCCL), which extends GPU-parallel computing tasks to the Quantum-X800 network fabric, taking advantage of its powerful in-network computing capabilities with SHARPv4 supporting FP8, performance upgrades for large model training, and generative AI.
NVIDIA’s full-stack software approach provides advanced programmability, making data center networks more flexible, reliable and responsive, ultimately increasing overall operational efficiency and supporting the needs of modern applications and services.
Ecosystem momentum
Next year, the Quantum-X800 and Spectrum-X800 will be available from a wide range of leading infrastructure and systems vendors worldwide, including Aivres, DDN, Technologies DellEviden, Hitachi Vantara, Hewlett Packard Enterprise, LenovoSupermicro and VAST Data.