🌊 GPU Computing in Oceanography on Google Cloud

Why this guide's resources? The whole tutorial is a chain of small cloud resources: a CloudBank billing account (federated login) opens the Google Cloud project, Cloud Shell drives gcloud, a Compute Engine GPU VM (1x T4) hosts the simulation, a firewall rule exposes Jupyter, and Jupyter or VS Code gives you access to the notebook. Read 0 - The Cloud Resources in This Tutorial first; the guides then follow the sequence 0 -> 4.

This showcase demonstrates the power of GPU-accelerated, differentiable ocean modeling using Julia. By leveraging Reactant.jl (XLA) and Enzyme.jl (Automatic Differentiation), we can simulate complex ocean dynamics and calculate sensitivities with unprecedented speed.

📁 Showcase Track: Interactive Exploration (Single GPU)

Learn the fundamentals of differentiable oceanography in an interactive Jupyter environment.

• Main File: ocean_gpu_tutorial.ipynb
• Shared Logic: ocean_utils.jl (Modular physics engine)
• Key Concepts: Ocean physics, topography (ridges), and calculating cause-and-effect (adjoint gradients).
• Setup Guides:
  • 0 - Cloud Resources Overview
  • 1 - Create a GPU Instance
  • 2 - Access via Browser
  • 3 - Access via VS Code
  • 4 - Execute the Tutorial Notebook

---

🚀 Getting Started

If you are new to this showcase, start with Step 1: Create a GPU Instance to set up your environment on Google Cloud.