Ori Global Cloud has unveiled a new Private Cloud cluster powered by Nvidia #00 GPUs, designed specifically for AI workloads of a large scale. This state-of-the-art cluster comprises 1,024 Nvidia #00 GPUs, organized according to the Nvidia DGX SuperPOD reference architecture. The formidable setup includes 128 Nvidia HGX nodes and is bolstered by one petabyte of Weka parallel storage, which enables high-speed read and write operations.
Technical Specifications and Performance
The technical specifications and performance metrics of this new Private Cloud cluster are truly impressive. The nodes are interconnected with a 3.2 Tbps Infiniband GPU interconnect, and they feature storage connectivity of 400 Gbps. During rigorous testing phases, the cluster consistently demonstrated sustained performance of 32 petaflops over several days. This substantial figure highlights that it operated at approximately 95 percent of its theoretical peak capacity, showcasing its robust capabilities.
CEO Insights and Future Prospects
Mahdi Yahya, the CEO of Ori Industries, emphasized that the new Private Cloud is tailored to equip AI-driven businesses with state-of-the-art technology. This ensures not only high performance and scalability but also top-notch security. Yahya also mentioned that the cluster is set to offer not just Nvidia #00s but will soon introduce access to Nvidia ##00s and possibly even Blackwell GPUs. This move underscores Ori’s commitment to remaining at the cutting edge of technological advancements.
Company Background and Global Operations
Ori Industries, founded in 2018 and headquartered in London, UK, has kept the exact deployment location of the new cluster under wraps. Nonetheless, the company has previously collaborated with various data center providers such as EdgeConneX in Europe and Centersquare in the US. This highlights Ori’s extensive global operations, with a footprint spanning 20 locations worldwide.
Industry Trends and Comparisons
Ori Global Cloud has introduced a cutting-edge Private Cloud cluster tailored for expansive AI workloads, leveraging the power of Nvidia #00 GPUs. This advanced cluster consists of an impressive 1,024 Nvidia #00 GPUs, meticulously arranged based on the Nvidia DGX SuperPOD reference architecture. Underpinning this powerful system are 128 Nvidia HGX nodes, providing a robust foundation for intensive AI computations. Complementing this framework is one petabyte of Weka parallel storage, ensuring rapid read and write operations vital for large-scale AI tasks.
This new offering from Ori Global Cloud epitomizes a significant leap in AI infrastructure, providing the necessary horsepower to handle the increasing demands of AI research, development, and deployment. The integration of Nvidia’s top-tier GPUs ensures optimal performance, while the DGX SuperPOD architecture offers scalability and efficiency. The substantial Weka storage allows for seamless data handling, an essential requirement for AI workloads that depend on fast and reliable data access. This makes Ori Global Cloud’s Private Cloud cluster a compelling choice for organizations looking to advance their AI capabilities.
