Skip to main content
Accelerating Your Discoveries

At Supermicro, we take pride in building HPC solutions from the ground up. From custom design to implementation, our dedicated teams offer plug-and-play services optimizing every aspect of each solution to meet and exceed even the most daunting challenges in HPC, while also simplifying the deployment of supercomputers.

General Purpose Compute Intensive

General Purpose Compute Intensive

Scientific Research Supercomputer

Scientific Research Supercomputer

AI/ML/DL GPU Intensive

AI/ML/DL GPU Intensive

HPC Solutions

Supermicro Reference Architecture is designed to fulfill your unique HPC requirements. Our advantages include a wide range of building blocks, from motherboard design, to system configuration, fully integrated rack, and liquid cooling systems. We focus on providing solutions tailored to the customers’ specific needs.

Enterprise HPC Reference Design

Built for general purpose compute intensive workloads, this versatile reference architecture is ideal for enterprise HPC use cases such as digital manufacturing, financial risk management, medical research, oil and gas exploration, and more.

  • Servers: SuperBlade® SBI-420P-1C2N, for up to 80 nodes per rack
  • Up to 6400 processor cores (Intel® Xeon®), 320TB DDR4-3200, and 160 2.5” drive bays per rack
  • Reduced cabling with built-in IB HDR and 25G switch modules
  • Good serviceability with hot-pluggable nodes, switch modules, PSU and FAN modules
  • Highly scalable, up to 10 racks and 800 compute nodes per cluster in a non-blocking IB fabric

Scientific Research Supercomputer Reference Design

Proven HPC design built for numerous scientific research projects, this supercomputer design is highly scalable for weather simulation, nuclear reaction physics simulation, gene sequencing, earth and space discovery, and more.

  • Servers: e.g. 16 TwinPro® SYS-220TP-HTTR or BigTwin® AS -2124BT-HTR, 64 nodes per rack
  • Up to 8192 cores with AMD EPYC™ (BigTwin) or 5120 cores with Intel Xeon per rack (TwinPro)
  • Optional DLC liquid cooling with 40KW CDU up to 80KW
  • Extremely scalable, up to 30 racks and 1920 nodes per cluster in a 2:1 OPA fabric
  • Balanced cost per performance with TwinPro’s one socket per node design, or uncompromised performance with BigTwin’s dual socket per node.

AI, Deep Learning HPC Reference Design

Packed with the latest breakthrough AI capabilities as they rapidly emerge, our scalable AI + HPC plug-and-play design is unlocking unprecedented speed of scientific discoveries and allowing an enterprise to focus on building game-changing AI applications for autonomous driving, natural language processing, and more.

  • Servers: e.g. 5 4U 8-GPU systems, SYS-420GP-TNAR or AS -4124GO-NART, with NVIDIA® HGX A100 and NVLink®
  • Up to 80 NVIDIA A100 GPUs with 40 Petaflops of AI performance per rack
  • Support up to 40TB DDR4-3200 and 30 hot-swappable Gen4 NVMe drivers per rack
  • High speed GPU interconnects with 8 200G IB HDR + 2 100G Ethernet ports per node
  • Highly scalable, up to 100 nodes and 800 A100 GPUs per cluster in a non-blocking IB HDR fabric with optional D2C liquid cooling

Rack Integration and Deployment Services

A series of progressive steps to ensure technologies and products meet customer’s unique HPC requirements with smooth deployment.

1. Design

  • Application Analysis
  • Power Budget
  • BOM Creation
  • Rack Layout

2. Assembly

  • Node Assembly
  • Rack & Stack
  • Cabling & Labeling
  • Networking & Power

3. Configuration

  • BIOS Setting Check
  • Firmware Management
  • Switch & IP Addresses
  • OS & Customer Image

4. Testing

  • Multi-vendor Interoperability
  • Full Rack Burn-in & QA
  • Full Rack Test Report
  • Performance Report

5. Logistics

  • Asset Labeling & Docs
  • Crating
  • Air-Ride Trucking
  • White Glove Services
Success Cases

Supermicro can tailor HPC solutions to meet any variety of workloads: compute intensive, high throughput GPUs, or high capacity storage applications used in different industries. Supermicro HPC solutions can be bundled with various open-source platforms and commercial applications with proven successes.

Osaka University, “The SQUID”

Osaka University worked closely with NEC Corporation and Supermicro to design and implement a new supercomputing class system based on a combination of Supermicro SuperBlade® systems and 4U 8-GPU NVIDIA NVlink® systems, which is built with the 3rd Gen Intel Xeon Scalable Processors.  The SQUID system is a 27 Rack cluster that contains 1520 SuperBlades with over 120,000 cores, 42 nodes of the4U 8 GPU NVIDIA NVLink systems, and was integrated and tested by the Supermicro Rack Integration Services team.

IMS, “The Molecular Simulator”

Japan’s Institute for Molecular Science (IMS) has installed two clusters called the High Performance Molecular Simulator with Supermicro servers interconnected by Intel Omni-Path Architecture. The Molecular Simulator achieves 1.8 petaFLOPS Linpack and 3.1 petaFlops theoretical peak performance through 40588 cores of Intel Xeon processors and 216,768GB of memory, running molecular dynamics’ massively parallel computations and many quantum chemistry’s more demanding serial operations.

LLNL, “Corona”

Lawrence Livermore National Laboratory (LLNL) selected Supermicro to provide additional computing capacity and resources for scientists working to find a cure for COVID-19. Supermicro’s 4U 8 GPU servers for this plug-and-play cluster equipped with nearly 1,000 AMD Radeon Instinct™ graphics accelerators and, combined with AMD’s state-of-the-art EPYC™ CPUs, expands the cluster to deliver up to 11 Petaflops of computing power for advanced computational workloads.