Skip to main content

Olga — a managed HPC cluster for research.

Quantum simulation, machine learning, and data-heavy workloads for ARISE affiliates, academic partners and vetted public-interest projects. Scheduled with Slurm, provisioned with Spack, and backed by a small engineering team that actually picks up the phone.

512
CPU cores
48
GPU accelerators
12TB
Memory
99.94%
Uptime (90d)
11
Partner institutions
23
Active projects

From request to first job — usually inside a week.

Step through the flow. Click any hotspot or use the arrow keys.

Step 1 / 4

Submit a request

Fill the short intake form with your project summary and resource estimate. Academic and public-interest applications are prioritised.

Hardware & software

A concise fact sheet. For deeper questions (driver versions, compiler toolchains, MPI library builds), talk to the engineering team via the contact form.

  • Compute nodes
    64 × dual-socket EPYC 7713 (64 cores / 2.0 GHz)
  • Accelerators
    48 × NVIDIA H100 80 GB SXM5 (across 6 DGX-style nodes)
  • Memory per node
    512 GB DDR4 ECC
  • Interconnect
    HDR InfiniBand 200 Gb/s · non-blocking fat tree
  • Parallel storage
    1.2 PB Lustre (all-NVMe) + 4 PB cold archive
  • Scheduler
    Slurm 23.x · fair-share + partition QoS
  • Software stack
    Spack · Apptainer/Singularity · EasyBuild modules
  • Data transfer
    Globus endpoint · dedicated 10 Gbps public uplink

Before you apply.

  • Researchers, students and engineers affiliated with academic or public-interest institutions. Private-sector access is handled case-by-case with our operations team.

Tell us what you want to run.

Three short steps. We'll respond within a few days — usually sooner.

We use this to verify academic / public-interest affiliation.