Ready for What’s Next: Exascale

For many years, Cambridge University, Dell Technologies and Intel Corporation have worked together to advance high performance computing (HPC). In joint collaborations, our HPC and AI experts test new cutting-edge technologies, design new systems and collaborate with the community to keep pace with a constantly evolving IT landscape.

Today, experts from Dell, Intel and Cambridge are collaborating to make exascale technologies more widely available. These efforts, under the umbrella of the new Cambridge Open Exascale Lab, are dedicated to facilitating innovation in emerging exascale technologies. The Lab enables technology leaders and experts from around the world to explore, test, and advance next-generation technologies.

So what is exascale, and why does it matter?

Today, the fastest supercomputers can run petascale workloads, performing a quadrillion calculations each second. Exascale systems are 100 to 1,000 times faster than current HPC systems. Exascale systems can perform a quintillion calculations per second, which equals a billion calculations in the blink of an eye.

While the computational power of exascale is mind-boggling, this is about more than systems with millions of cores processing massive amounts of data. It’s really about enabling scientists to solve some of the world’s most pressing problems, from climate change and environmental pollution to deadly diseases that trigger pandemics. These are really big problems that can’t be solved without massive amounts of compute and storage, powerful analytics and AI tools that reach beyond the limits of the human brain.

Exascale systems can help us meet these and a host of other challenges. Computer simulations — such as modeling global weather patterns, assessing drug effectiveness and designing new clean energy processes — require huge amounts of processing power. Scientific software must scale across many tens of thousands of server cores to deliver results with enough speed and detail to generate accurate, meaningful information within the time required to solve real-world problems. The Cambridge Open Exascale Lab will help us get there, sooner rather than later.

Five key projects define the initial focus for the Cambridge Open Exascale Lab. A regular series of papers and technology updates will detail progress of these projects, with additions in the future.

• oneAPI Centre of Excellence — oneAPI is a new scalable programming model from Intel that targets heterogenous systems and eases the task of programming for, and running code on, different compute architectures. The oneAPI Centre of Excellence will focus on adapting key existing scientific applications for the exascale era, future-proofing the code and speeding scientific discovery.

• Scientific OpenStack — Scientific OpenStack creates a very powerful and flexible environment to deploy, manage and use large-scale HPC systems. It supports onsite, cloud-based and hybrid deployments, allowing scientists and researchers to deploy clusters on-demand to run their particular workflows while taking advantage of the leading technologies in exascale platforms.

• Exascale-class storage solutions — The lab will investigate a range of file system technologies deployed on solid-state storage devices. Initial work will focus on Intel’s open-source DAOS object store, which is designed from the ground up to deliver extreme-scale HPC storage performance.

• High-performance Ethernet for exascale — The lab will investigate multi-vendor low-latency Ethernet for HPC solutions and benchmark a diverse range of HPC applications at a range of scales. The lab will also investigate software-defined Ethernet technologies for HPC and stimulate cross-industry focus on open standards for high-performance Ethernet.

• Extreme scale visualization — The lab will use large-scale solid-state storage test beds combined with large-scale rendering compute engines and leading HPC graphical software to demonstrate real-time network rendering of extreme-scale datasets. These techniques will be documented and packaged to provide tangible tools that are simple to use and straightforward to implement.

Ultimately, through projects like these, the Cambridge Open Exascale Lab will help ensure that tomorrow’s supercomputers will be able to take on the challenges that await them — and that this vitally needed technology will be accessible for the many, and not just the few.

To Learn More

Get involved. To learn how to access the resources of the Cambridge Open Exascale Lab, as well as other labs and Dell Technologies Centers of Excellence, contact the University of Cambridge, Dell Technologies, Intel or an authorized partner sales representative.

Join the Dell Technologies HPC Community at dellhpc.org and explore more HPC solutions Dell Technologies and Intel.