Skip to main content

About Research Computing

ASU Research Computing provides high-performance computing (HPC) infrastructure and expertise to support academic research across disciplines. These computing resources include access to multi-node clusters with thousands of CPU cores, hundreds of GPUs, and other specialized hardware.

These systems are available to ASU faculty, staff, students, and approved collaborators. Services are designed to enable large-scale data analysis, machine learning, simulations, modeling, and visualization.

ASU Research Computing offers two public clusters and a high-sensitivity HIPAA-compliant cluster.

When Should I Use a Supercomputer?

Use Supercomputing resources if:

  • Your workload exceeds local workstation capabilities
  • You require long runtimes or large-scale parallelization
  • You work with large data sets or simulations
  • You perform machine learning, deep learning, or GPU-based computing
  • You require access to specialized hardware (e.g., high-memory nodes or GPUs)

Supercomputer Specs​

Sol Supercomputer​

  • 21,000+ CPU cores using the modern AMD EPYC architecture
  • Nodes with 512GB of RAM
  • Over 200 GPUs (including A100, 80GB A100s and 80GB H100s)
  • Seven high memory nodes with 2TB of memory
  • ARM-based Grace Hopper Nodes
  • FPGA hardware from Xilinx and Bittware

See full list of Sol hardware.

Phoenix Supercomputer​

  • 17,000+ CPU cores using heterogeneous Intel-based architecture
  • Nodes with 128GB to 384GB of RAM
  • 360 GPUs (including NVIDIA A100, V100, GTX 1080, and RTX 2080)
  • Three high memory nodes with 1.5TB of memory

See full list of Phoenix hardware.

Aloe Supercomputer​

The Aloe supercomputer is part of the KE Secure Cloud environment. KE Secure Cloud is designed for sensitive and regulated data types, such as HIPAA and FERPA data. For more information about the KE Secure Cloud environment, check out the KE Secure Cloud documentation.

  • 1,700+ CPU cores using the modern AMD Epyc architecture
  • 20 A100 80GB GPUs
  • 2 High Memory nodes with 1TB RAM

Cost Structure​

SystemCostNotes
SolNo CostResource allocation is dynamically adjusted by users' recent historical usage
PhoenixNo CostLast-generation hardware well-suited for highly-paralleziable workloads
AloePaidRequires KE Secure Cloud access and data review