Browser Guide

The RunMat sandbox is a browser-based development environment for writing and running MATLAB-style code with GPU acceleration. It provides a full IDE experience—code editor, file explorer, console output, and live plotting—all running locally in your browser.

Sandbox vs Sandbox + Cloud

What is the RunMat sandbox?

The sandbox lets you:

• Write MATLAB-style code in a full-featured editor with syntax highlighting
• Run scripts instantly with automatic CPU/GPU acceleration
• See live plots rendered directly in the browser
• Inspect variables and console output in real time

All computation happens locally in your browser. There's no server-side execution—your code and data never leave your machine.

Accessing the sandbox

Visit runmat.com/sandbox to launch the sandbox.

No installation required. Works in Chrome, Edge, Firefox, and Safari.

Note: For GPU acceleration, use a browser that supports WebGPU (Chrome 113+, Edge 113+, Safari 18+, Firefox 139+). Resource limits may vary by browser—some browsers are more conservative to preserve battery life.

The Interface

When you open the sandbox, you'll see three main areas:

┌─────────────┬────────────────────────┬─────────────────────┐
│   Sidebar   │        Editor          │   Runtime Panel     │
│  (Files)    │  (Code + Tabs)         │  (Run, Output, Vars)│
└─────────────┴────────────────────────┴─────────────────────┘

Sidebar (Left)

• File tree showing your project files
• Click files to open them in the editor
• Use the + button to create new files or folders

Editor (Center)

• Full code editor with syntax highlighting and language services (e.g., red underlines for errors)
• Multiple file tabs (click a file to open, double-click to pin)
• Unsaved changes shown with a dot indicator
• Cmd/Ctrl+S to save

Runtime Panel (Right)

• Run button to execute the current file
• Figure tabs for viewing plots
• Console for standard output and input prompts
• Variables pane showing workspace variables, their types, shapes, and whether they're on CPU or GPU

All three panels are resizable—drag the borders to adjust.

First Run Walkthrough

When you first open the sandbox, a set of example files is already loaded.

Step 1: Click "Run demo.m"

In the Runtime Panel on the right, click the purple ▶ Run demo.m button.

Step 2: View the Output

After execution:

• The Console shows the printed output
• The Variables tab (at the bottom of the Runtime Panel) displays all workspace variables with their classes, shapes, and residency (CPU/GPU)

Step 3: Modify and Re-run

Try editing the script. For example, add a plot:

plot(a, b);
title("Sine Wave");

Click Run again. A new Figure tab appears in the Runtime Panel showing your plot.

Creating Your Own Scripts

Create a New File

1. Click the + button in the sidebar
2. Select Create file (.m)
3. Enter a filename (e.g., my_script.m)
4. Press Enter

Write Your Code

The editor supports standard MATLAB syntax:

% Element-wise operations (automatically GPU-accelerated for large arrays)
x = rand(1000000, 1);
y = sin(x) .* exp(-x);

% Built-in functions
avg = mean(y);
fprintf('Average: %.6f\n', avg);

% Plotting
plot(x(1:1000), y(1:1000));
xlabel('x');
ylabel('sin(x) * exp(-x)');

Run Your Script

• Click Run in the Runtime Panel, or
• Use the keyboard shortcut Ctrl+Enter (Windows/Linux) or Cmd+Enter (macOS)

How It Works

Local Execution

The sandbox runs entirely in your browser using WebAssembly. When you click Run:

1. Your script is compiled and executed locally
2. RunMat automatically decides whether to use CPU or GPU for each operation
3. Results stream to the console and variable inspector in real time

GPU Acceleration

For large arrays and intensive math, RunMat automatically fuses operations and runs them on your GPU (via WebGPU). This happens transparently—you don't need to write any special GPU code. Browsers throttle GPU and CPU usage to preserve battery life, so sandbox benchmarks will be slower than the CLI running the same code natively. For full GPU throughput, use the CLI.

The Variables pane shows Residency for each variable:

• cpu — data is on the CPU
• gpu — data is on the GPU (faster for large operations)

Plotting

Plots render directly in the browser using GPU-accelerated graphics. Currently supported:

• plot — line plots
• scatter — scatter plots
• surf, mesh — 3D surface plots (interactive rotate, zoom, pan)

Additional chart types (bar, histogram, subplots, figure handles) are in progress.

Storage

Without an account (sandbox mode)

Without signing in, the sandbox runs entirely in your browser. Your files are stored in memory within your browser tab, and the RunMat runtime executes locally via WebAssembly. Your code stays on your machine—it is never sent to our servers.

• No account required
• Your code stays local
• Files are cleared when you close or refresh the tab

With a RunMat Cloud account

Sign in to get cloud storage, automatic file versioning, and access to your projects from any device. Your files sync to RunMat Cloud and persist across sessions.

• Hobby tier — 100 MB storage, unlimited projects, automatic version history
• Pro — 10 GB storage, version history ($30/mo per user)
• Team — 100 GB storage, project sharing, SSO ($100/mo per user)

File versioning

Every time you save a file, RunMat Cloud records a version automatically. You can browse and restore previous versions of any file at any time. Version history is available on all Cloud tiers (Hobby, Pro, and Team). Stored versions count toward your storage quota.

When signed in, your code is transmitted to RunMat Cloud for storage and sync. Execution still happens locally in your browser—your code is not executed on our servers.

See pricing for full plan details. For local persistence without an account, use the CLI.

Common Questions

Why don't I see GPU acceleration?

GPU acceleration requires:

• A browser that supports WebGPU (Chrome 113+, Edge 113+, Safari 18+, Firefox 139+)
• A compatible GPU with up-to-date drivers

If WebGPU isn't available, RunMat falls back to CPU execution. You can still run all the same scripts—they just won't get GPU speedups.

Can I use my existing MATLAB files?

Yes. RunMat supports standard .m file syntax. Many MATLAB scripts run with few or no changes.

See the Language Coverage guide for supported features.

How do I handle user input?

Use input() for interactive prompts:

name = input('Enter your name: ', 's');
fprintf('Hello, %s!\n', name);

When your script calls input(), the console prompts you for input.

What if I want full desktop performance?

Browsers limit how much GPU and CPU a website can use to preserve battery life. For maximum performance:

1. Install the RunMat CLI — Run scripts from your terminal with full native GPU access. See the CLI guide.
2. RunMat desktop app — Coming soon. Will provide a native IDE with full local file system access.

Teams and Enterprise

Team features (Cloud Team plan)

• Project sharing — share projects with collaborators in your organization. Each editor is billed as a seat.
• Team workspaces — organize projects under your team's organization.
• SSO / SAML — single sign-on for your identity provider.
• Priority support — faster response times from the RunMat team.

Enterprise

For organizations that need on-premises or air-gapped deployment:

• Self-hosted deployment — run RunMat on your own infrastructure.
• Data residency and ITAR compliance — keep data in your environment.
• Audit logs — track access and changes for compliance.
• Offline licensing — no internet connection required.
• SCIM provisioning — automated user management.

See pricing for plan details. For Enterprise inquiries, contact sales or email team@runmat.com.

What's Next

Now that you've run your first script:

• Explore the built-in functions — See the Function Reference
• Learn about GPU acceleration — Read GPU Acceleration
• Install the CLI — For local file access and scripting: CLI Guide
• Try the benchmarks — Compare RunMat performance against NumPy and PyTorch

Getting Help

• Documentation: runmat.com/docs
• GitHub Issues: Report bugs or request features
• Discussions: Ask questions and share ideas