Tiny libraries. Massive performance. Runs anywhere.
forKernels is a portable ecosystem of high-performance computing kernels that fit on any device and plug into any language. From a Raspberry Pi to a Jetson to a cloud VM — same library, same API, same speed.
Each module is a single lightweight native library — typically under 2 MB — with a clean C ABI that works with any language or runtime that can call a function pointer.
Python. C++. Rust. C#. JavaScript. Swift. Go. Java. Anything.
No frameworks. No runtimes. No dependencies. Just drop in a library and call it.
| ✅ | Device |
|---|---|
| 🤖 | NVIDIA Jetson Orin Nano / Thor |
| 🍓 | Raspberry Pi 5 |
| 🕶️ | Wearables & smart glasses |
| 💻 | Laptops & desktops (x86, ARM) |
| ☁️ | Cloud VMs (AWS, GCP) |
| 🌐 | Browsers (WASM) |
One library. Every platform. No rewrite.
17+ specialized kernel libraries spanning:
| 🧠 AI & ML | 📷 Computer Vision | 🎮 3D & Simulation | 📡 Signal Processing |
| 📈 Financial Computing | 🧬 Bioinformatics | 🧭 Navigation & SLAM | 🎲 Bayesian Inference |
| ⚡ Optimizers | 🔢 Linear Algebra | 📊 FFT & Transforms | 💾 Async I/O |
Use one module or all of them. They're independent, composable, and designed to work together.
| Benchmark | Result |
|---|---|
| 100K particle fluid simulation | 122 FPS on edge hardware |
| Financial time-series (SMA) | 1,962× faster than pandas |
| 3D mesh transforms | 16–24× faster than JS |
| Real-time CV pipeline | All targets met on Jetson Orin Nano |
Available on AWS Marketplace and Google Cloud Marketplace.
| Tier | |
|---|---|
| Free | Core math and FFT kernels |
| Pro | All modules, Python bindings |
| Enterprise | Source access, custom deployment, priority support |
Device licensing available for embedded, robotics, and edge deployment.
forkernels.com · hello@thefantasticplanet.com
Published by The Fantastic Planet