HPE Cray supercomputer to boost Singapore’s met office weather forecasting

HPE has built a new Cray supercomputer for the Meteorological Service Singapore (MSS) to advance weather forecasting and tropical climate research for Singapore and the broader Southeast Asia region.

The new hardware is set to replace MSS’s existing system, delivering nearly twice as much performance and advanced capabilities across compute, storage, software, and networking.

It will be hosted at MSS’s data centre in the Centre for Climate Research Singapore (CCRS). The organisations underlined that as climate modelling becomes increasingly complex due to new seasonal patterns, it requires faster, more sophisticated supercomputing capabilities.

The supercomputer will help increase forecast skills through improved numerical model configurations and advanced data assimilation of ground-and space-based weather observations in MSS’s numerical weather prediction system called ‘SINGV’.

Additionally, it will enable the use of modern forecast postprocessing algorithms, using machine learning techniques, to improve the quality of forecasts.

HPE said the enhanced computational power will help the MSS to further develop models to deliver improved weather and climate products. This includes sub-kilometre scale urban modelling (uSINGV), which is being developed to represent the urban environment more appropriately in weather and climate applications.

An improved coupled ocean-atmosphere-land-wave modelling system (cSINGV) will also capture and improve the understanding of the strong feedbacks between the atmosphere, land and ocean, which have a significant impact on the weather and climate over the Southeast Asia region.

“At CCRS, our scientists and software engineers are committed to developing advanced modelling systems and examining complex data to provide timely weather forecasts for our nation, which due to the island’s unique geological positioning, often experiences various of weather processes on a daily basis,” said Prof. Dale Barker, director at CCRS.

“After collaborating with Hewlett Packard Enterprise to design the new supercomputer, our research centre will gain a faster system with next-generation technologies to advance modelling and simulation tools, while introducing new capabilities to test and apply future types of applications for deeper research methodologies.”

MSS will get nearly double the performance, with a peak performance of 401.4 teraflops. It has 196 3rd-gen AMD EPYC processors to target modelling and simulation of weather and climate data as well as HPE Slingshot to deliver an additional boost in speed while enabling congestion control for larger data-intensive and AI workloads.

The supercomputer contains a Cray ClusterStor E1000 storage parallel system and a framework to address storage demands of data growth and complexity from an increase of workloads in modelling, simulation, and AI.

Lastly, the new system will continue to use Altair PBS Professional, a job-scheduling and workload manager to efficiently scale and optimise various modelling, simulation, and AI workloads.

With a peak performance of 401.4 teraflops, it is significantly slower than other supercomputers HPE has built recently. In December 2021, the company was set to build a supercomputer for Thailand’s science agency, with a peak performance of 13 petaflops. A petaflop is equal to 1,000 teraflops.

Additionally, in September 2022, HPE revealed it would build a system for a university in Saudi Arabia, which is set to reach 100 petaflop/s of performance. It’s set to deliver supercomputing and artificial intelligence capabilities for research in fields like food, water, energy, and the environment.