ASPIRE2A’s core computing capabilities deliver a level of performance and flexibility needed to support a multifaceted array of HPC applications. The computational components are balanced with high speed storage subsystems and a low latency high speed interconnect that ensures to deliver the highest levels of performance across a broad spectrum of applications.

ASPIRE2A is an AMD-Based Cray EX supercomputer with 25 PB of GPFS FS and 10 PB of Lustre FS storage and Slingshot interconnect. The following diagram shows the high-level specification of ASPIRE2A.

The NSCC HPC system comprises of the following:

ASPIRE2A

  • HPE Cray EX 2x AMD EPYC Millan 7713 providing total compute capacity of up to 10 PFlops, 512 GB memory and 128 cores per node.
  • GPU compute capability with 4 x NVIDIA A100-40G SXM per node.
  • Please click here for more info

AI System

  • Total 18 AI GPU nodes
    • 12 nodes with 4x Nvidia A100 40GB and 12 TB nvme local storage.
    • 6 nodes with 8x Nvidia A100 40GB and 14TB nvme local storage.
    • Access to the AI systems is via ASPIRE2A “ai” queue.

High Frequency

  • 16 DL385 High Frequency Nodes.
    • Dual-CPU AMD 75F3 (32 cores/CPU + 32 cores/CPU = 64 cores in a node).
    • 100G High speed network.
    • 512GB DDR4 ECC RAM (User accessible RAM = 500 GB).
    • Red Hat Enterprise Linux-8

High Speed Network Interconnect (HPE Slingshot)

  • All nodes are connected with HPE Slingshot Interconnect (Dragonfly Topology).
  • HPE Slingshot provides a modern, high-performance interconnect for HPC and AI clusters that delivers high-bandwidth and low-latency for HPC, ML, and analytics applications.

Other Features

  • Remote extended network connections to the A*STAR, NUS, NTU, SUTD and NEA sites.
  • Parallel file system (Lustre & PFSS)
  • Liquid cooled high-density Cray EX cabinets.
  • Air cooled racks (specialized AI, Large memory, storage, login nodes)
  • Altair Workload Manager

 

Technical Specifications click here

 

Interested to find out more about ASPIRE 2A? Get in touch with us at [email protected].