Powering the exascale era with enterprise accelerated computing
This revolution is driven by the need for intelligence, which is inspiring enterprises to leverage their data in strategic ways that will transform how they work, compete, and grow. Cutting-edge technologies like high-performance computing (HPC) and artificial intelligence (AI) are enabling a major shift, delivering critical insights to fuel innovation and solve the world’s greatest problems. However, today’s enterprises are being overwhelmed by a deluge of data and struggle to unlock its intelligence.
The digital universe is expanding at an unprecedented rate. By 2025, IDC predicts that worldwide data will grow 61%, from 45 zettabytes in 2019 to 175 zettabytes. With this enormous increase, HPC and AI tasks are erupting in complexity. Welcome to the age of exascale.
Exascale marks the beginning of a new era of discovery which demands a more sophisticated approach to computing. Every industry faces the challenge of converting vast amounts of data into real-time insights through the use of modeling, simulation, data analytics, and IoT. These complex workloads are converging, creating workflows that exceed the capabilities of traditional supercomputing. In order to succeed, enterprises must invest in the next generation of technology with greater power and acceleration than ever before.
Computing innovations to propel discovery
Digital transformation is the cornerstone of the exascale era. Yet exascale is more than a system size or speed milestone. It’s the emergence of advanced workloads triggered by the joint evolution of science and computing. The ability to harness deeper insight is key to remain competitive, from honing products and services to quickly executing rigorous HPC and AI tasks. To meet these escalating requirements, enterprises must integrate the latest technologies to modernize their environments and operate at extreme scale.
Accelerated computing is expanding the use of HPC and AI for enterprise. This groundbreaking development empowers organizations to drastically reduce time-to-insight and time-to-market by providing superior performance with faster, more reliable processing capacity. Everyone can benefit — from small enterprises wanting to realize better outcomes, to the largest labs researching climate change and cures for cancer.
Designing and implementing a new compute infrastructure can be daunting for organizations that lack the expertise or resources. That’s why we’re introducing a breed of accelerated computing to help enterprises of every size and scope achieve exascale.
Proven solutions for the exascale era
HPE combines immense compute power with the unparalleled speed of GPUs to optimize the most demanding workloads. Today, our breakthrough solutions are taking these capabilities to the next level.
For example, the HPE Apollo 6500 Gen10 Plus System is designed for GPU-intensive processes that suit AI, machine learning and deep learning while also executing typical HPC workloads such as modeling and simulation. They support up to 16 GPUs. The platform supports the new AMD Instinct MI100 GPU with Infinity Fabric Link. Powered by the first AMD Compute DNA architecture, MI100 GPUs deliver a significant leap in compute and connectivity, offering nearly 3X the performance for HPC and nearly 6X the performance for AI workloads, compared to the prior generations of AMD accelerators.
The system also features the high core count of AMD EPYC 7002 series processors. High-frequency AMD processors integrated with HDR InfiniBand add up to 200 gigabytes per second of bandwidth for every two GPUs, allowing enterprises to communicate at twice the speed. These solutions are perfectly aligned to optimize today’s tasks and prepare for tomorrow’s exascale challenges.
Pankaj Goyal is Vice President of HPE’s AI business. Visit HPE online to learn more about HPE’s enterprise products.