Overview
Each cell in Conway's Game of Life only needs its eight neighbours to determine its next state, which makes the simulation a textbook case for GPU acceleration. This project implements the next-generation step as a CUDA kernel — one thread per cell — and benchmarks it against a baseline serial CPU implementation across grid sizes, CUDA block sizes, and visualization update rates.
Highlights
- Up to 131× speedup over the serial implementation on a GTX 1080
- Studied block-size and visualization-frequency impact on throughput
- Per-iteration pixel comparison to verify GPU output matches CPU output
- OpenCV used for grid visualization and image output
Tech Stack
CUDA
C++
OpenCV
NVIDIA GPU