Supermicro Announces Broad Portfolio of New GPU Systems

Supermicro, Inc., a manufacturer of Total IT Solutions for AI, Cloud, Storage, and 5G/Edge, has unveiled a wide range of new GPU systems based on the NVIDIA reference architecture. These systems feature the latest NVIDIA GH200 Grace Hopper and NVIDIA Grace CPU Superchip. The modular architecture is designed to standardize AI infrastructure and accelerated computing in compact form factors, providing flexibility and expansion options for GPUs, DPUs, and CPUs. Supermicro's liquid-cooling technology allows for high-density configurations, such as a 1U 2-node setup with integrated NVIDIA GH200 Grace Hopper Superchips. The company can deliver thousands of AI servers per month and ensures Plug-and-Play compatibility.

"Supermicro is a leader in driving the AI revolution and transforming data centers to support AI workloads," said Charles Liang, president and CEO of Supermicro. "Our highly modular and scalable systems are optimized for rapidly evolving AI technologies. By collaborating with NVIDIA, we are accelerating time to market for enterprises and simplifying deployment while reducing environmental impact. Our new servers incorporate the latest industry technology, including NVIDIA GH200 Grace Hopper Superchips, BlueField, and PCIe 5.0 EDSFF slots."

"NVIDIA and Supermicro have a long history of collaboration on high-performance AI systems," said Ian Buck, vice president of hyperscale and HPC at NVIDIA. "The NVIDIA MGX modular reference design, combined with Supermicro's server expertise, will create new generations of AI systems that benefit customers and industries worldwide."

Supermicro's NVIDIA MGX Platform Overview

Supermicro's NVIDIA MGX platforms are designed to accommodate future AI technologies. The product line addresses the unique challenges of AI-based servers in terms of thermal, power, and mechanical requirements.

The new NVIDIA MGX line of servers from Supermicro includes:

  • ARS-111GL-NHR - 1 NVIDIA GH200 Grace Hopper Superchip, Air-Cooled
  • ARS-111GL-NHR-LCC - 1 NVIDIA GH200 Grace Hopper Superchip, Liquid-Cooled
  • ARS-111GL-DHNR-LCC - 2 NVIDIA GH200 Grace Hopper Superchips, 2 Nodes, Liquid-Cooled
  • ARS-121L-DNR - 2 NVIDIA Grace CPU Superchips in each of 2 Nodes, 288 Cores in total
  • ARS-221GL-NR - 1 NVIDIA Grace CPU Superchip in 2U
  • SYS-221GE-NR - Dual-socket 4th Gen Intel Xeon Scalable processors with up to 4 NVIDIA H100 Tensor Core or 4 NVIDIA PCIe GPUs

Each MGX platform can be enhanced with NVIDIA BlueField-3 DPU and/or NVIDIA ConnectX-7 interconnects for high-performance networking.

Technical Specifications

Supermicro's 1U NVIDIA MGX systems feature up to 2 NVIDIA GH200 Grace Hopper Superchips with 2 NVIDIA H100 GPUs and 2 NVIDIA Grace CPUs. They come with 480 GB LPDDR5X memory for the CPU and 96 GB of HBM3 or 144 GB of HBM3e memory for the GPU. The memory-coherent NVIDIA-C2C interconnects the CPU, GPU, and memory at 900 GB/s, which is 7 times faster than PCIe 5.0. The modular architecture provides multiple PCIe 5.0 x16 FHFL slots for DPUs, additional GPUs, networking, and storage.

Supermicro's 1U 2-node design with 2 NVIDIA GH200 Grace Hopper Superchips utilizes direct-to-chip liquid cooling solutions, reducing OPEX by over 40% and increasing computing density for large language model clusters and HPC applications.

The 2U Supermicro NVIDIA MGX platform supports both NVIDIA Grace and x86 CPUs with up to 4 full-size data center GPUs. It also provides PCIe 5.0 x16 slots for I/O connectivity and hot-swap EDSFF storage bays.

Supermicro offers NVIDIA networking solutions, including BlueField-3 DPUs and ConnectX-7 adapters, to secure and accelerate AI workloads on the MGX platform.

Developers can quickly utilize these new systems and NVIDIA software offerings for various industry workloads. NVIDIA AI Enterprise software powers the NVIDIA AI platform and streamlines the development and deployment of generative AI, computer vision, speech AI, and more. The NVIDIA HPC software development kit provides essential tools for scientific computing.

Supermicro's NVIDIA MGX systems are designed for efficiency, with intelligent thermal design and component selection. NVIDIA Grace Superchip CPUs deliver up to 2x the performance per watt compared to standard x86 CPUs. Specific Supermicro NVIDIA MGX systems can be configured with two nodes in 1U, providing groundbreaking compute densities and energy efficiency in hyperscale and edge data centers.