Samsung and Intel: Partners in the Age of Supercomputing

Supercomputing (or high-performance computing) was once the stuff of science fiction, and now it’s everywhere, even if most of us don’t realize it. Samsung Semiconductor CSVP of Memory Sales Jim Elliott sat down with Intel’s VP and GM of AI, HPC, and Datacenter Accelerators Solutions and Sales Nash Palaniswamy to discuss supercomputing and its increased importance in today’s day and age. The conversation began with an analysis of the evolution of the High Performance Computing (HPC) market, new applications for HPC, and how storage and memory plays a crucial role in these processes. “HPC is really a combination of two things,” said Palaniswamy, “a high-performing infrastructure, and the high-demand applications” supported by the infrastructure. As industries find more practical applications for AI, the need for powerful compute infrastructure to handle it increases exponentially. HPC infrastructure faces the challenge of being outpaced by AI advancements. When asked about examples of supercomputing’s contributions to life on our planet, Palaniswamy listed 3 examples of fields that benefit from HPC today: 1. Cryogenic electron microscopy (Cryo-EM) technology, an exciting field that won the 2017 Nobel Prize in Chemistry. Cryo-EM utilizes HPC to make three-dimensional models of viruses to help identify them and their variants – useful for breaking down the structure of the COVID-19 virus. 2. Improved brain mapping for neuroscientists and researchers relies on HPC to advance both medical science and AI modeling. 3. HPC-supported breakthroughs in fusion energy discoveries that aid in safe, efficient reactor management. As complex applications like AI and machine learning have grown, the high-performance memory and storage capabilities to empower these advances (and handle the data associated with them) have evolved in tandem. “I believe storage is the unsung hero of HPC and innovation in this area,” Palaniswamy said. “(Memory and Storage) keep the HPC systems humming, and make it possible to leverage all the data.” Palaniswamy then mentioned the Samsung DDR5 and DRAM memory portfolio as direct contributors to the innovation and advancements in supercomputing technology—particularly how much DDR5 contributes to better data transfer performance over the last generation (DDR4). “You couldn’t do that before. You had to go break [the data] into multiple [packets]. With advancements in DDR5, you can do it in one go—you have better precision. With DDR5, you build better products.”

Elliott and Palaniswamy then dove into exascale computing. They discussed the difficulties of achieving the quintillion-calculations-per-second level of computing, and the challenges of architecting resilient and reliable computing infrastructure, using open standards. Again, DDR5 along with its memory bandwidth and efficiency scaling were credited as pivotal to this next-generation computing process. The two capped off their conversation by recognizing the significance of the collaborative partnership between Samsung and Intel in HPC. “Our two companies share the same vision of making things possible,” Palaniswamy said. “To enrich the lives of people on this planet.” Watch the full interview: