Champollion: Bringing HPC & purpose-built AI technologies to train AI at scale

[Nabanita1]

Organizations are embracing Artificial Intelligence (AI) to harness insights and improve business and research outcomes, and with this, High performance computing (HPC) has become the backbone to many AI initiatives that require large computational force to train models at scale, in a short period of time, to unlock results faster.

With the help of HPC and purpose-built AI technologies, HPE is fueling capabilities to speed up AI projects. Demonstrating this is the new AI supercomputer the company is building, in collaboration with NVIDIA, called “Champollion.” Champollion was hosted in the HPE Center of Excellence (CoE) in Grenoble, France, where they offered HPC and AI resources to our partners, customers and broader community to test and optimize their solutions for market.

Champollion marks the latest milestone in AI innovation to improve intelligence in products and services to make better decisions, faster. The new AI supercomputer will be built using 20 HPE Apollo 6500 Gen10 Plus systems, which deliver an end-to-end HPC and AI platform, featuring 160 NVIDIA A100 Tensor Core GPUs for accelerated compute to target AI model development and training, and NVIDIA Quantum InfiniBand for networking.

Additionally, by combining the HPE Machine Learning Development Environment, an optimized software stack that is ready for users to easily develop and train AI models faster, at scale, we are removing the need to purchase, deploy and tune technologies required to develop and train models.

Follow us on Twitter | Join our LinkedIn group | Read OEM Solutions blog on the HPE Community | Register here