Eviden, a company within the Atos group and a leader in advanced computing, has announced the launch of JARVICE™ AI, an innovative software solution designed to streamline the work of data scientists and AI infrastructure administrators. This new product complements its full line of BullSequana AI products and services, providing a comprehensive solution for orchestrating geographically dispersed GPU clusters and cloud providers.
An Extension of JARVICE™ XE
JARVICE AI is positioned as an advanced extension of Eviden’s enterprise HPC platform, JARVICE™ XE. This new tool introduces specific AI capabilities, allowing one-click orchestration of MLOps frameworks on advanced infrastructures, whether on-premises, in the cloud, or in federated architectures. It offers bare metal performance and flexibility to run a variety of AI acceleration resources, including GPU, CPU, and IPU.
Innovations and Benefits
The solution leverages the extensive HyperHub™ application marketplace, expanding it with a catalog of AI-specific applications and frameworks in areas such as generative AI, model operations, and large-scale language models (LLM). Additionally, JARVICE AI supports custom AI applications and frameworks, enabling organizations to tailor to their specific needs.
One of the standout features of JARVICE AI is its integrated advanced workload scheduler. This scheduler can manage a diverse set of concurrent AI scheduling needs from a single platform, including large-scale training for generative AI frameworks, low-latency inference tasks, and complex MLOps workflows. This functionality ensures high levels of cluster utilization efficiency and workload mobility.
Enhanced User Experience
“By combining and simplifying the management and deployment of AI and HPC applications under a single control panel across all organization compute endpoints, JARVICE AI provides a secure and scalable user experience for scientists, researchers, and engineers,” commented Leo Reiter, CTO of Nimbix Cloud. “With JARVICE AI and HyperHub, organizations can confidently and easily meet the rapidly evolving AI workload demands, no matter what the future brings.”
Tanmay Bakshi, ML architect and software at IBM Automation, and Google Developer for Machine Learning at IBM, also expressed excitement: “JARVICE AI excites me as a user of JARVICE for 6 years for large-scale deep learning model training. It’s user-friendly, unifies MLOps workflows across any infrastructure in a single control panel, and introduces new advanced workload scheduling – it’s a very compelling offering for the enterprise.”
Availability and Launch
JARVICE AI will be available starting in the third quarter of 2024, although customers can already access it in advance. Eviden unveiled this innovative solution at ISC24, the high-performance computing event held in Hamburg, Germany.
With JARVICE AI, Eviden strengthens its position as a leader in advanced computing and provides companies with a powerful tool to enhance the efficiency and capability of their AI operations.