November 12, 2025
Hot plates, hotter takes
Simulating a Planet on the GPU: Part 1 (2022)
Solo dev tries to build SimEarth 2—fans hype it, skeptics say “you need a supercomputer”
TLDR: A solo dev spent a year simulating a planet on the graphics card, from plates to sand-like crust. Commenters split between hype and 'needs a supercomputer,' linked 'proper' tectonics guides, asked how far GPUs go, and mourned the missing Part 2—ambition vs reality in one thread.
Remember SimEarth? One fearless coder got tired of waiting and tried to build a whole planet from scratch—on the graphics card—so you can watch continents crash, oceans swirl, and mountains rise. The community went wild. The shader crowd swooned, calling it “coding from the perspective of a pixel” and cheering the compute-shader pivot (that’s using the GPU—your graphics card—to crunch math in parallel). Nostalgia? Off the charts. Someone joked about writing to Will Wright, and another quipped, “1990 called, it wants its planet back.”
Then came the reality check crew. The top hot take: you’ll need a team and a supercomputer to do this for real. The “proper simulation” folks swooped in with tectonics.js, sparking a nerdy turf war: good-enough planet vs scientifically correct Earth. Curious onlookers asked the most relatable question: “How much can a GPU even do? Like, how many points can you move?”
But the heartbreak twist stole the thread: there’s no Part 2—life happened. Cue memes about “Schrödinger’s sequel” and “coming soon (probably).” Fans still applauded the sand-like physics idea (think falling-sand games, but for crust, air, and water), and crowned the dev a one-person disaster movie. Verdict: big dreams, bigger drama, and hot plates with even hotter takes.
Key Points
- •The author aims to build a realistic planet simulation inspired by SimEarth, focusing on deep processes like plate tectonics.
- •Early polygon-based methods using Delaunay and Voronoi in Unity/C# hit performance limits, especially for realistic plate collisions.
- •A brief shift to C++ with SDL2 and Vulkan highlighted the complexity of writing a custom engine, leading back to Unity.
- •GPU compute shaders and cubemap representations enabled plate collisions, subduction, and seafloor spreading but lacked crust deformation.
- •Smoothed-Particle Hydrodynamics is being considered to achieve realistic deformation of crust and potentially model air and water.