GPU Accelerated Diamond-Square
See the article: https://www.vertexfragment.com/ramblings/diamond-square/
This repository presents a GPU accelerated implementation of the classic Diamond-Square algorithm, alongside a multi-threaded C# approach. As shown in the table below, the GPU version is 17x faster than the standard single-threaded approach and 4x faster than the multi-threaded implementation which runs a worker for every 128x128 chunk.
All timings are in milliseconds.
| Implementation | 512 x 512 | 1024 x 1024 | 2048 x 2048 | 4096 x 4096 | 8192 x 8192 |
|---|---|---|---|---|---|
| Single-threaded C# | 38.3 | 161.6 | 740.2 | 2987.3 | 13042.1 |
| Multi-threaded C# | 26.7 | 67.9 | 224.4 | 823.9 | 3121.5 |
| GPU-Accelerated | 2.1 | 9.3 | 41.7 | 211.6 | 778.2 |
Benchmarked on an Intel i7-4790K and NVIDIA GeForce GTX 980 Ti.
Requirements
Unity v2022.1.11f or later to run the demo.
The multi-threaded implementation (DiamondSquare.cs) can be used outside of Unity, however the GPU implementation (DiamondSquareGPU.cs) uses the Unity Compute Shader API but the shader itself can be plugged into other engines.