Success Stories
xAI Colossus
- 100,000 Accelerators (NVIDIA Hopper Tensor Cores GPUs)
- NVIDIA H100/H200 GPUs
- 1:1 networking to each GPU for node cluster with advanced all-flash NVMe for fastest AI data pipeline
- Large Scale AI Training applications
- Serve The Home Article
Cooling Type - Air Cooling & Liquid to Liquid (Direct Liquid Cooling)
Success Story
University of Tsukuba – Pegasus
- 120 Supermicro SuperBlades® in 4th Gen Intel® Xeon® Scalable Processors
- NVIDIA H100 GPU
- Huge memory space with Intel Optane Persistant Memory 300
- HPC and AI applications
- Ranked #12 on Green500 List and #190 on Top500 List in June 2023
Success Story
Osaka University – SQUID
- 1,520 general-purpose CPU nodes (3rd Gen Intel Xeon Scalable Processors)
- 42 GPU nodes (NVIDIA A100)
- Supermicro SuperBlades
- Ranked #57 on Green500 List and #67 on Top500 List in June 2021
Cooling Type – Liquid to Liquid (Direct Liquid Cooling)
Success Story
Lawrence Livermore National Laboratory “Ruby”
- COVID-19 Research
- 1,528 nodes (Intel Xeon) with Intel Deep Learning Boost
- Ranked #79 on November 2020 Top500 List
Cooling Type – Liquid to Liquid (Direct Liquid Cooling)
Success Story
Supermicro Rack Scale Solutions Enable Faster Time to Productivity With Reduced TCO
xAI Colossus (from @ServeTheHome)
One of the most powerful liquid-cooled AI SuperClusters is designed to take xAI’s Grok AI to the next level. Supermicro is developing leading liquid cooling technology powering the world’s most ambitious AI infrastructure projects.
Rack Scale Testing Facilities with AMD and NVIDIA
Insight of Supermicro rack scale integration testing facilities from design to deployment, and to see what offers in L11 and L12 testing and validation
Rack Scale Solution with Intel Processors
Supermicro is ready to rack! Supermicro’s industry leading rack integration service and rack production capacity, capable for building rack-scale solutions for customers based on new 4th Gen Intel® Xeon® Scalable processors
Rack Plug and Play Cloud Infrastructure
Supermicro’s Rack Scale Plug and Play Solutions eliminate the burden of designing and deploying a custom in-house cloud solution with a pre-configured, pre-validated rack-level total configuration
Why Partner with Supermicro for Rack Scale Solutions
Minimize Rack Deployment Costs
- Receive Fully-Tested L12 Clusters
- Rely on Cluster-Level Strenuous Burn-In Tests
- Plug and Play Design Enables Customers to Plug In, Connect to Network and Liquid Cooling, and Become Productive Sooner
- Customer Software Installed and Tested
Maximize Supermicro Factory Expertise and Delivery Times
- Design Entire Cluster with Supermicro Hardware, Software, Networking and Power Experts
- Collaborate on Alternative Designs
- Reduce Power Expenditure With a Range of Application-Optimized Servers
- Worldwide Manufacturing of 5,000 Racks/Month
Optimize At Rack Scale
- Determine Best Platform and Components with Supermicro Product Managers
- Select Optimum Power Supplies for Workloads and Power Budget
- Architect Network Topology and Switches Together with Experts
- Optimize Server, Storage, and Switch Placement
- Take advantage of a full turn-key AI data center solution, SuperCluster, for mission-critical enterprise use cases, eliminating the complexity of building a large cluster
Acquire a Total IT Solution
- Design, Build, Test, and Deliver to Customer Sites
- Software Installed and Benchmarked to Customer Requirements
- System Management Software and Tracking Enabled Upon Request
7 Phases of the Rack Integration Process
| Rack Design |
|
|---|---|
| Quoting & Confirmation |
|
| Assembly |
|
| Burn-In & Testing |
|
| Delivery |
|
| On-Site Deployment |
|
| Signature Support |
|
Rack Scale Solutions
Accelerated Time-To-Delivery of Plug and Play Racks with Proven Quality and Reliability