Penguin Computing, a leader in high-performance computing (HPC) and artificial intelligence (AI), will offer customized designs for Intel Select Solutions for HPC and AI Converged Clusters. Penguin Computing, a long-time partner for Intel-based HPC systems, joined the Intel Select Solutions ecosystem in early 2020.
“Increasingly, industries are utilizing the powerful combination of HPC and AI to solve their difficult data-driven problems,” Matt Jacobs, Chief Strategy Officer of Penguin Computing explained. “They need these new HPC and AI converged architectures to use machine learning to narrow the scope of the questions they need to ask of their modeling and simulation runs. With these new converged solutions, organizations are able to run both types of workloads in a single platform with technologies optimized for both types of computing.”
Penguin Computing has been building customized, complex, built-to-order systems for nearly two decades. They deliver some of the fastest HPC systems in the world to leading scientific and commercial enterprises. Intel Select Solutions for HPC and AI Converged Clusters gives them a baseline architecture to continue to build their customized systems.
Intel Select Solutions are workload-optimized reference designs that minimize the challenges system architects face when building solutions for their infrastructure. This reference architecture comprises tailored combinations of Intel compute, memory, storage, and network technologies that are optimized to increase the value of hardware and applications investments. The solutions are validated and certified by industry builders, such as Penguin Computing, and verified by Intel. The Intel Select Solution for HPC and AI Converged Clusters addresses an emerging use case.
“This demand for convergence is largely driven by new customers coming into the market who need these tools to design innovative products,” Jacobs said. “But they don’t necessarily have the same understanding of these technologies that long-time practitioners do. It’s a new generation of scientists and researchers finding unique ways to solve their problems with converged HPC and AI. So, they need these solutions to be easily adopted to accelerate their time to solution and productivity.”
But technology evolution continuously introduces new approaches and tools to solve tomorrow’s scientific problems. Understanding new CPU capabilities, model frameworks, and software-defined architectures, among others, introduces new challenges and growing complexity to building optimal computational platforms. A baseline architecture helps make new solutions more ‘easily adopted.’
“It’s all about reducing complexity,” Jacobs added. “The scientific domains we deal with have quite different requirements. The hardware and software-defined technologies to address their needs continue to get more complex, and the approaches to build balanced compute, network, and data storage systems to solve their higher science and data problems can be quite varied. Starting with a proven reference architecture helps reduce complexity and increase availability of these platforms for customers who don’t necessarily have a lot of technical expertise. A baseline architecture accelerates time to market and the customer’s time to solution.”
According to Jacobs, having a reference architecture also eases the challenges many industries, including HPC solution builders, face with supply chains these days. As the complexity of systems increases, acquiring the best parts to evaluate becomes more difficult. Using the Intel Select Solution for HPC and AI Converged Clusters helps Penguin Computing ease those supply chain challenges with a validated, verified architecture of a reduced number of components.
“We’ve been working with Intel programs like this for a while,” Jacobs concluded. “With Intel Select Solutions, the ability to build more systems off fewer part numbers gives us greater supply chain efficiency, which allows us to deliver more solutions faster to customers. Additionally, these engagements provide customers access to the Intel teams to help on the engineering side and provides context and homogeneity in a highly configurable technology environment.”
Intel technologies may require enabled hardware, software or service activation.
No product or component can be absolutely secure.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.