A million times more powerful than a laptop, exascale supercomputers could help solve today’s most pressing problems that much faster. However, they’re too complex to be programmed and used by most scientists. Watch Dr. Paul Calleja, Director of Research Computing at the University of Cambridge, describe the promises and challenges of the dawning exascale and zettascale computing era.
At a glance:
- The Cambridge Open Zettascale Lab looks at the pain points of next-generation, high-performance computers.
- Designed for heterogeneous computing, Intel® oneAPI is an open, unified, and cross-architecture programming model.
Heterogeneous computing—which includes exascale and zettascale systems—can dramatically improve the computing power of a given system. It uses different types of processors to solve complex mathematical tasks, breaking down larger tasks into smaller ones and assigning them to the different processors available.
Towards exascale and zettascale systems
Most things created today have been modelled in supercomputers. This goes for everything from personal electronics to medicine to nuclear weapons. As exascale computing becomes a reality and zettascale becomes possible, the promise for scientists, engineers, and clinicians to tackle global-scale challenges is closer than ever, but certain challenges must be overcome.
“Exascale and zettascale computing could solve a lot of ‘grand challenge’ problems,” says Dr. Paul Calleja, Director of Research Computing Services and the Cambridge Open Zettascale Lab, University of Cambridge. “The computation provided by such systems is a big step forward compared to most systems today—but those systems have different pain points […].”
Intel oneAPI: Democratizing exascale and zettascale solutions
Exascale supercomputers have the computational power of approximately one million laptops all working together on a single problem. They’re very challenging due to their sheer complexity and the power they consume. These systems aren’t easy to program, and scientists and engineers can’t simply sit down and start to use them. If they’re unable to program them, they can’t use them to solve today’s problems.
Part of the solution—according to the Cambridge Open Zettascale Lab—is to use technologies such as Intel® oneAPI . Intel oneAPI is a new programming environment focused on portability and scalability of codes on large-scale supercomputers, which makes accessing this class of systems much easier. It provides on-demand cloud-like access to these incredibly complicated supercomputers.
As Robert Maskell, Director of High-Performance Computing, Modelling and Simulation at Intel explains: “Intel oneAPI is [the key to] getting people to think about open programming. So, when they're writing a program, they can write it once and it runs on many different architectures.”
Wider challenges for the high-performance computing (HPC) future
The move towards zettascale computing continues, with the announcement that the 1.1 exaflop Frontier supercomputer at Oak Ridge National Laboratory demonstrated performance of more than 10^18 (1 quintillion) operations per second in standardized tests . Not far behind, engineers at the US Department of Energy (DOE) Argonne National Laboratory are building the Aurora supercomputer, an exascale system based on Intel technologies .
While the challenges present in the world of exascale are considerable, with solutions like Intel oneAPI, these systems will enable scientists, engineers, and clinicians in almost all walks of life to tackle global-scale challenges—from understanding the universe to designing new materials and developing personalized medicine and healthcare.