Want to Watch Evolution Happen in Real Time? This CUDA-Powered Artificial Life Sim Lets You Do Exactly That
You've probably played with evolution simulators before—little blobs wiggling around, maybe some basic genetic algorithms. But what if you could simulate millions of particles in real time, each one part of an organism controlled by a neural network, all running on your GPU? That's what ALIEN (Artificial Life Environment) is trying to pull off.
What It Does
ALIEN is an artificial life simulation tool built on a specialized 2D particle engine written entirely in CUDA. The core idea is straightforward: each simulated body is a network of particles, and those networks can be upgraded with higher-level functions—sensors, muscles, weapons, constructors, and so on. These functions are orchestrated by neural networks, which means the bodies operate as agents or digital organisms in a shared environment.
The bodies aren't just random collections of particles, either. Their blueprints can be stored in genomes and passed on to offspring, with cell-by-cell construction of those offspring during reproduction. The physics engine handles soft and rigid body mechanics, fluids, heat dissipation, damage, and adhesion. Rendering happens through OpenGL using CUDA-OpenGL interoperability, so everything stays on the GPU.
The project is optimized for large-scale real-time simulations with millions of particles. It supports NVIDIA GPUs (compute capability 7.5 or higher—GeForce RTX 20 series and up) and AMD GPUs (RDNA2 or newer). There's also a headless Docker image for running on cloud instances.
Why It's Cool
-
It's a complete ecosystem, not just a physics demo. The combination of particle physics, neural network control, and a genetic system means you can actually watch evolution happen. Turn on mutations, let self-replicating machines loose, and the simulation does the rest. That's a rare thing to see working in real time.
-
The editing tools are surprisingly deep. There's a graph editor for manipulating every particle and connection, freehand and geometric drawing tools, and a genetic editor for designing custom organisms. You're not locked into whatever the simulation spawns—you can build your own creatures from scratch.
-
It's built for scale. Writing the simulation entirely in CUDA isn't just a flex. It means you can run simulations with millions of particles without your framerate tanking. The CUDA-OpenGL interoperability keeps rendering fast too, so you're not waiting on data transfers between CPU and GPU.
-
The networking features are a nice touch. There's a built-in simulation browser where you can download and upload simulation files, and you can upvote simulations by giving stars. It's a small thing, but it turns the project into more of a community rather than just a solo sandbox.
-
The project has a clear research goal. The developer is upfront about wanting to better understand the conditions for pre-biotic evolution and the growing complexity of biological systems. This isn't just a toy—it's a tool for asking real questions about how life might emerge from simple rules.
-
It won the ALIFE 2024 Virtual Creatures Competition. The demo video, "Emerging Ecosystems," took first place. That's not nothing—it suggests the simulation is capable of producing genuinely interesting emergent behavior.
How to Try It
Getting started depends on your platform and GPU.
-
Check your hardware. You'll need an NVIDIA GPU with compute capability 7.5 or higher (RTX 20 series or newer) or an AMD GPU with RDNA2 or newer.
-
On Windows, you can grab a nightly build or build from sources. The same goes for AMD GPUs on Windows.
-
On Linux, you'll need to build from sources for both NVIDIA and AMD.
-
For cloud instances, there's a Docker image available for NVIDIA GPUs (headless mode). AMD isn't supported in the cloud yet.
-
Clone the repository and follow the build instructions:
https://github.com/chrxh/alien -
Join the Discord if you run into issues or want to discuss the project. There's an active community around ALIEN and artificial life in general.
Final Thoughts
ALIEN isn't for everyone. If you don't have a compatible GPU, you're out of luck—there's no CPU fallback. And building from sources on Linux isn't exactly a one-click install. But if you're interested in artificial life, evolutionary simulation, or just want to see what millions of particles can do when you give them neural networks and genomes, this is one of the more ambitious projects out there.
It's best suited for researchers, hobbyists with decent hardware, and anyone who's ever wanted to watch evolution unfold in real time rather than reading about it in a paper. The fact that it's open source and actively developed makes it worth checking out—even if you just want to poke around the code and see how someone built a CUDA-based particle engine from scratch.
Follow @githubprojects for more developer tools and open source projects.