Huawei’s Already Using ARM Supercomputers

At the recent Supercomputing Conference, Huawei showed off its own high-performance computing platform (HPC). It also introduced its Atlas 900 AI Cluster. It is used in tandem with HPC to serve a variety of applications that include domains, supercomputing centers, manufacturing simulation, weather prediction, gene sequencing, and energy exploration.

The Atlas 900 AI Cluster and Huawei’s ARM based HPC form its supercomputing prototype, used for astronomical data analysis at the SKA regional center in China. The company says that its ARM HPC uses TaiShan 200 servers as its primary compute and management nodes.

Huawei says that it works with “Atlas’ heterogeneous nodes and x86 nodes processes the complex workloads of cosmic reionization, spectral lines, continuous spectrum imaging, pulsar searches, galaxy classification, and transient source searches.”

Independent shared storage and task scheduling are done with data islands, which are interconnected through high-speed networks to flexibly reorganize resources. This promotes flexibility and stability and allow it to process large sets of astronomical data. Huawei is heavily invested in the HPC domain, which it hopes can provide diversified puting power for the world.