Getting Started with RunMat

Get up and running with RunMat in minutes. This guide will walk you through installation, basic usage, and your first interactive session.

Installation

Linux & macOS

curl -fsSL https://runmat.org/install.sh | sh

Windows

iwr https://runmat.org/install.ps1 | iex

More Installation Options

Your First RunMat Session

1Start the Interactive REPL

Open your terminal and start RunMat:

$ runmat

RunMat v0.0.1 by Dystr (https://dystr.com)

High-performance MATLAB/Octave runtime with JIT compilation and GC

JIT compiler: enabled (Cranelift optimization level: Speed)

Garbage collector: "default"

No snapshot loaded - standard library will be compiled on demand

Type 'help' for help, 'exit' to quit, '.info' for system information

runmat>

2Try Basic Calculations

Start with simple arithmetic and variables:

runmat>

x = 5

ans = 5

runmat>

y = 3.14

ans = 3.14

runmat>

result = x * y + 2

ans = 17.7

3Work with Matrices

Create and manipulate matrices using familiar MATLAB syntax:

runmat>

A = [1, 2, 3; 4, 5, 6]

ans = [1 2 3; 4 5 6]

runmat>

B = A * 2

ans = [2 4 6; 8 10 12]

runmat>

C = A + B

ans = [3 6 9; 12 15 18]

4Experience Automatic GPU Acceleration

RunMat keeps data on the GPU and fuses operations automatically—no device flags required.

runmat>

x = rand(1000, 1000); % Million elements

runmat>

y = sin(x) .* exp(x); % Fused on GPU automatically

runmat>

m = mean(y, 'all'); % Stays on GPU

[GPU] Fused 2 operations, 1M elements in 0.8ms

5Create Your First Plot (Pre-release)

Generate simple 2D line or scatter plots. Advanced plot types are still in progress, so expect rough edges while we expand coverage.

runmat>

x = [0, 1, 2, 3, 4, 5]

ans = [0 1 2 3 4 5]

runmat>

y = [0, 1, 4, 9, 16, 25]

ans = [0 1 4 9 16 25]

runmat>

plot(x, y)

[Pre-release plot window opens for 2D line/scatter]

Basic zoom/pan controls ship today; rotate and richer tooling arrive with future plotting milestones.

Running MATLAB Scripts

Execute .m Files

Run existing MATLAB/Octave scripts directly with automatic GPU acceleration:

# Run a script file

runmat script.m

# Run with specific options

runmat run --jit-threshold 100 simulation.m

Most MATLAB and GNU Octave scripts will run without modification. Check our compatibility guide for details.

Jupyter Notebook Integration

1Install RunMat as a Jupyter Kernel

Make RunMat available as a kernel in Jupyter notebooks:

runmat --install-kernel

RunMat Jupyter kernel installed successfully!

Kernel directory: ~/.local/share/jupyter/kernels/runmat

One-time setup that works with existing Jupyter installations

2Start Jupyter and Select RunMat

Launch Jupyter and create notebooks with the RunMat kernel:

# Start Jupyter Notebook

jupyter notebook

# Or Jupyter Lab

jupyter lab

# Then select "RunMat" when creating a new notebook

Full MATLAB syntax support with automatic GPU acceleration

3Verify Installation

Check that the RunMat kernel is properly installed:

jupyter kernelspec list

Available kernels:

python3 /usr/local/share/jupyter/kernels/python3

runmat ~/.local/share/jupyter/kernels/runmat

If you don't see RunMat listed, ensure Jupyter is installed and try running the install command again.

Next Steps

Learn the Fundamentals

Dive deeper into RunMat's features and capabilities.

How RunMat Works

Explore Examples

See RunMat in action with real-world examples.

Benchmarks

Discover RunMat on the GPU

How RunMat turns MATLAB scripts into GPU-accelerated workloads

RunMat Accelerate

Need Help?

Join our community and get support from other RunMat users and developers.

GitHub Discussions Report Issues