# Selene

> supercomputer developed by Nvidia

**Wikidata**: [Q107394659](https://www.wikidata.org/wiki/Q107394659)  
**Wikipedia**: [English](https://en.wikipedia.org/wiki/Selene_(supercomputer))  
**Source**: https://4ort.xyz/entity/selene-q107394659

## Summary
Selene is a supercomputer developed by Nvidia, launched in 2020. It is ranked among the world's most powerful supercomputers and uses Nvidia's DGX A100 systems with AMD EPYC 7742 processors and NVIDIA A100 GPUs.

## Key Facts
- Developed by Nvidia, an American multinational technology company founded in 1993
- Uses AMD EPYC 7742 64C processors running at 2.25GHz
- Contains 555,520 processor cores
- Achieves 63.46 petaflops performance on LINPACK benchmarks
- Consumes 2,646 kilowatts of power
- Ranked 6th on the TOP500 list as of November 2021
- Connected via Infiniband HDR networking
- Runs Ubuntu 20.04.1 LTS operating system
- Has a TOP500 system ID of 179842
- Located in the United States

## FAQs
### Q: What is Selene used for?
A: Selene is used for high-performance computing tasks including AI research, scientific simulations, and data analysis. It serves as a research and development platform for Nvidia and its partners.

### Q: How powerful is Selene compared to other supercomputers?
A: As of November 2021, Selene ranked 6th on the TOP500 list of the world's most powerful supercomputers, achieving 63.46 petaflops of performance.

### Q: What hardware components make up Selene?
A: Selene consists of Nvidia DGX A100 systems, AMD EPYC 7742 64C processors, NVIDIA A100 GPUs, and Mellanox HDR Infiniband networking components.

## Why It Matters
Selene represents a significant advancement in AI and high-performance computing infrastructure. As one of the world's most powerful supercomputers, it enables breakthrough research in artificial intelligence, machine learning, and scientific computing. The system demonstrates Nvidia's capability to build large-scale computing systems that push the boundaries of what's possible in computational research. Its performance and architecture serve as a reference point for the industry, influencing the design of future supercomputers and data center infrastructure. Selene's development also showcases the growing importance of GPU-accelerated computing in solving complex computational problems that were previously intractable.

## Notable For
- Ranked among the top 15 supercomputers globally since its deployment
- One of the first large-scale deployments of Nvidia's DGX A100 systems
- Achieves industry-leading performance-per-watt efficiency
- Serves as a development platform for next-generation AI technologies
- Demonstrates the scalability of GPU-accelerated computing architectures

## Body
### Architecture and Specifications
Selene is built using Nvidia's DGX A100 systems, which integrate multiple components into a single computing unit. The system employs AMD EPYC 7742 64C processors as its CPU foundation, providing 64 cores per processor running at 2.25GHz. The GPU layer consists of NVIDIA A100 accelerators, which are specifically designed for AI and high-performance computing workloads.

### Performance Metrics
The supercomputer achieves 63.46 petaflops of sustained performance on LINPACK benchmarks, which is the standard metric for ranking supercomputers on the TOP500 list. Its theoretical peak performance reaches 79.22 petaflops. These performance levels place Selene among the elite computing systems globally, capable of processing quadrillions of operations per second.

### Infrastructure and Connectivity
Selene utilizes Mellanox HDR Infiniband networking for high-speed inter-node communication, essential for distributed computing workloads. The system consumes 2,646 kilowatts of power during operation, reflecting the significant energy requirements of modern supercomputing infrastructure. The computing nodes are interconnected through a high-bandwidth, low-latency network fabric that enables efficient parallel processing.

### Software and Operating Environment
The system runs Ubuntu 20.04.1 LTS as its operating system, providing a stable and well-supported foundation for scientific computing workloads. This Linux-based environment supports the various software frameworks and libraries required for AI research, scientific simulations, and data analytics tasks that the supercomputer is designed to handle.

### Development and Deployment
Launched in 2020, Selene was developed during a period of rapid advancement in AI computing capabilities. The system represents Nvidia's vision for the future of computing, where GPU acceleration becomes central to solving complex computational problems. Its deployment marked a significant milestone in the commercialization of AI supercomputing technology, making capabilities previously available only in research laboratories accessible to a broader range of users and applications.

## References

1. [Source](https://top500.org/site/48448/)
2. [Source](https://www.top500.org/system/179842/)