# Eos NVIDIA DGX SuperPOD

> supercomputer released in 2023

**Wikidata**: [Q133584190](https://www.wikidata.org/wiki/Q133584190)  
**Source**: https://4ort.xyz/entity/eos-nvidia-dgx-superpod

## Summary
The Eos NVIDIA DGX SuperPOD is a supercomputer released in 2023 by Nvidia, designed for high-performance computing workloads. It features NVIDIA H100 GPUs and Intel Xeon Platinum processors, achieving significant computational power in artificial intelligence and scientific research.

## Key Facts
- Released in 2023 by Nvidia, an American semiconductor company headquartered in Santa Clara, California.
- Achieved a TOP500 ranking of 9th in November 2023, later ranking 10th (June 2024), 12th (November 2024), and 16th (June 2025).
- Equipped with 485,888 processor cores using Intel Xeon Platinum 8480C CPUs (56 cores at 3.8GHz) and NVIDIA H100 GPUs.
- Delivers measured performance (Rmax) of 121.40 petaflops and peak performance (Rpeak) of 188.65 petaflops.
- Uses InfiniBand NDR400 for high-speed interconnect between nodes.
- Operates on Ubuntu 22.04.3 LTS as its operating system.
- Classified as a supercomputer with TOP500 system ID 180239.
- Manufactured by Nvidia, leveraging its expertise in GPU technology for AI and HPC applications.

## FAQs
### Q: What is the primary purpose of the Eos NVIDIA DGX SuperPOD?
A: It is designed for high-performance computing tasks, particularly artificial intelligence training, scientific simulations, and large-scale data analysis.

### Q: How does it compare to other supercomputers?
A: It ranked among the top 16 globally in the TOP500 list from 2023-2025, demonstrating its position as a leading supercomputer in performance benchmarks.

### Q: What hardware components power this system?
A: The system combines Intel Xeon Platinum 8480C processors with NVIDIA H100 GPUs, connected via InfiniBand NDR400 networking.

### Q: What software does it use?
A: It runs Ubuntu 22.04.3 LTS as its operating system, optimized for Nvidia's GPU computing ecosystem.

## Why It Matters
The Eos NVIDIA DGX SuperPOD represents a significant advancement in scalable AI and scientific computing infrastructure. By integrating cutting-edge NVIDIA H100 GPUs with high-performance networking, it enables researchers and organizations to tackle complex problems like large-language model training and climate simulations that require massive computational resources. Its deployment underscores Nvidia's role in democratizing access to exascale-level computing capabilities for enterprise and scientific users. The system's dynamic TOP500 rankings also highlight the rapid evolution of supercomputing, where even top-tier systems require continuous upgrades to maintain competitiveness in an ever-accelerating field.

## Notable For
- Achieved TOP500's top-10 ranking within its first year of deployment (9th in November 2023).
- Combines NVIDIA H100 GPUs with Intel Xeon Platinum CPUs in a tightly integrated DGX architecture optimized for AI workloads.
- Utilizes InfiniBand NDR400 networking, providing high-speed interconnect crucial for scalable parallel processing.
- Features one of the highest core counts in its class (485,888 cores), enabling extreme parallelism for complex simulations.

## Body

### System Architecture
The Eos NVIDIA DGX SuperPOD is a pre-configured supercomputer pod manufactured by Nvidia in 2023. It operates as a high-performance computing (HPC) system designed for AI training and scientific workloads. The system integrates both CPU and GPU acceleration within a unified architecture, leveraging Nvidia's DGX platform for optimized software-hardware integration.

### Hardware Specifications
- **Processors**: Intel Xeon Platinum 8480C CPUs with 56 cores each, operating at 3.8GHz
- **Accelerators**: NVIDIA H100 GPUs for parallel computing tasks
- **Interconnect**: InfiniBand NDR400 for high-speed node-to-node communication
- **Core Count**: 485,888 total processor cores
- **Operating System**: Ubuntu 22.04.3 LTS

### Performance Metrics
- **Measured Performance (Rmax)**: 121.40 petaflops
- **Peak Performance (Rpeak)**: 188.65 petaflops
- **TOP500 Rankings**: 
  - 9th (November 2023)
  - 10th (June 2024)
  - 12th (November 2024)
  - 16th (June 2025)

### Manufacturer Context
Developed by Nvidia, a U.S.-based semiconductor company founded in 1993. The system represents Nvidia's expansion into integrated supercomputing solutions, combining its GPU expertise with high-performance infrastructure to serve AI research and enterprise computing needs. The DGX SuperPOD architecture is specifically engineered to minimize bottlenecks in large-scale AI model training and data-intensive simulations.

## References

1. [Source](https://top500.org/system/180239/)
2. [TOP500](https://top500.org/system/180239/)