Record breaking strength at  SC24

New Integrated Solution Simplifies AI Factory Buildout

Features Best in Class Technology from Eviden and VDURA to Accelerate Time to Value for AI Deployments While Reducing Cost and Complexity

By:   At the SC24 conference held here, SourceCode, a pioneer of hardware co-design from device to datacenter, today announced the Atlas AI Cluster, an integrated solution featuring technology from Eviden and VDURA to bring up AI factories with optimal cost and performance. Partnering with Eviden, a global leader in exascale-class supercomputing, and VDURA, whose recently-launched Data Platform software is purpose-built for AI storage needs, ensures customers receive the best-in-class solution for AI deployment.

(Atlas AI Cluster from SourceCode)
The Atlas solution is the fastest way to build an AI factory at the highest levels of performance and efficiency. The solution is available in four configurations, incorporating high-bandwidth, low-latency NVIDIA Quantum-2 400Gb/s InfiniBand networking, 120-208 NVIDIA GH200 Grace Hopper Superchips, and up to 13 PB of raw storage per rack. Atlas will be available exclusively through SourceCode in North America in early 2025.

The Atlas solution leverages a full suite of management and AI software, including NVIDIA NIM microservices, the NVIDIA AI Enterprise software platform, the NVIDIA Omniverse platform, and optimized open-source foundational models.

To support seamless integration, SourceCode will provide application assistance and optimization, knowledge transfer, cluster management software, and onsite integration. With the recent international expansion of SourceCode Labs, the company continues to build on its technical leadership in AI and HPC, creating AI hardware solutions that align with evolving software stacks and application needs.

As one of a handful of companies capable of engineering massive scale and hybrid supercomputing, Eviden sets the standard for flexibility, performance, and efficiency. The Atlas solution features Eviden exascale-class engineering, which are powering breakthroughs in AI, science, and engineering around the world. Atlas, along with Eviden’s world-class enterprise, HPC, AI, and quantum solutions, is available in North America through SourceCode, made possible by a landmark partnership between SourceCode and Eviden.

The new VDURA Data Platform software is designed to set new industry benchmarks in performance, scalability, and data management simplicity. Its inclusion in Atlas delivers AI-optimized metadata and input/output operations per second (IOPS), flexible deployment for maximum efficiency and simplicity, and enhanced data protection. IOPS is a measure of how fast a system can read and write data, an important metric for data-intensive AI applications.

“Across industries, customers are embracing the AI imperative,” said Ludovic Sauvage, Head of Sales, Advanced Computing at Eviden North America. “They need adaptable systems and software to confidently build and scale AI operations. By partnering with SourceCode, we’re excited to expand our reach, delivering state-of-the-art solutions to more markets and customers.”

“While the need for compute, interconnect, and software for AI is well-established, there is still a gap in the market when it comes to storage that is specifically optimized for AI workloads,” said Ken Claffey, CEO at VDURA. “That’s where Atlas comes in, providing the right mix of technologies and services to accelerate the deployment of AI factories for customers.”

As AI continues to revolutionize industries worldwide, SourceCode’s co-design model has emerged as a particularly effective approach for customized solutions.

“SourceCode’s business model is native to the requirements of today’s technology landscape: engineering-led, mission-focused co-creation, complemented by the most capable e-commerce platform for fulfilling ongoing needs,” said Arthur Ataie, CEO of SourceCode.

“Hands-on technical experience with the latest hardware and software gives our engineering team the nuanced insights that make all the difference for customer success,” said Matt Ritter, VP Engineering and Product at SourceCode. “We are ready to add value and support customers at any stage in their journey towards AI factories.”