GPU layout

For large graphs, the force layout can run on the GPU. Switch the renderer and engine and scale the node count to feel the difference:

Enabling it

Set useGPU on the force layout:

const orb = new OrbView(container, {
  render: { type: 'webgl' }, // pair with the WebGL renderer for large graphs
  layout: { type: 'force', options: { useGPU: true } },
});

It's the same force layout as the CPU version - the same forces and options apply - just computed on the GPU using WebGL2 and a Barnes-Hut approximation.

Automatic fallback

If WebGL2 isn't available, Orb silently falls back to the CPU force engine, so it's safe to request useGPU: true everywhere.

Trade-off: main thread

The CPU force layout runs in a web worker. The GPU layout needs a WebGL context, which a worker can't provide, so it runs on the main thread instead. For most large graphs the GPU speedup outweighs this; for small graphs there's no reason to leave the worker.

When to use it

Reach for the GPU layout once you're simulating many thousands of nodes and the CPU layout feels slow. Pair it with the WebGL renderer so both the layout and the drawing scale together.