Skip to content

ops-hpc

OpenSees HPC Pipeline Controller for Google Cloud

Run OpenSees parallel structural analyses on Google Cloud h3-standard-88 VMs (88 cores, Intel Sapphire Rapids). Manages the full lifecycle: build OpenSees image, run analyses, download results, clean up.

Features

  • GUI Dashboard — PySide6 interface with job queue, live output, VM metrics, panic buttons
  • CLI — scriptable commands for automation
  • Safety-first — VMs always deleted (quadruple safety net), cost tracking, crash recovery
  • Cloud Storage — results uploaded to GCS, auto-downloaded to your PC, storage freed automatically
  • Batch support — queue multiple analyses, pipelined uploads, per-job timeouts
  • Extend timeout — add time mid-analysis without restarting

Quick Start

pip install ".[gui]"
gcloud auth application-default login
gcloud compute project-info add-metadata --project=<YOUR_PROJECT_ID> --metadata enable-oslogin=TRUE
ops-hpc config create my-profile
ops-hpc setup
ops-hpc build
ops-hpc run --jobs jobs.txt

Costs (us-central1, April 2026)

Resource Cost
h3-standard-88 standard $4.92/hr
h3-standard-88 spot $2.34/hr
Image storage (50 GB) ~$2.50/mo
Image archived ~$0.65/mo
Empty bucket $0.00