Installation

System Requirements

Helion currently targets Linux systems and requires a recent Python and PyTorch environment:

Operating System

Linux-based OS
Other Unix-like systems may work but are not officially supported

Python Environment

Python 3.10, 3.11, or 3.12
We recommend using conda for environment management

Dependencies

PyTorch nightly build
Triton development version installed from source

Note: Older versions may work, but will lack support for features like TMA on Hopper/Blackwell GPUs and may exhibit lower performance.

Installation Methods

Method 1: Install via pip

The easiest way to install Helion is directly from GitHub:

pip install git+https://github.com/pytorch-labs/helion.git

We also publish PyPI releases, but the GitHub version is recommended for the latest features and fixes.

Method 2: Development Installation

For development purposes or if you want to modify Helion:

# Clone the repository
git clone https://github.com/pytorch-labs/helion.git
cd helion

# Install in editable mode with development dependencies
pip install -e '.[dev]'

This installs Helion in “editable” mode so that changes to the source code take effect without needing to reinstall.

Step-by-Step Setup Guide

1. Set Up Conda Environment

We recommend using conda to manage dependencies:

# Create a new environment
conda create -n helion python=3.12
conda activate helion

2. Install PyTorch Nightly

Install the latest PyTorch nightly build:

# For CUDA 12.6 systems
pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu126

see PyTorch installation instructions for other options.

3. Install Triton from Source

Helion requires a development version of Triton:

# Clone and install Triton
git clone https://github.com/triton-lang/triton.git
cd triton

# Install Triton
pip install -e .

# Return to your working directory
cd ..

4. Install Helion

Choose one of the installation methods above:

# Option A: From GitHub
pip install git+https://github.com/pytorch-labs/helion.git

# Option B: Development installation
git clone https://github.com/pytorch-labs/helion.git
cd helion
pip install -e '.[dev]'

Verification

Verify your installation by running a simple test:

import torch
import helion
import helion.language as hl

@helion.kernel(use_default_config=True)
def test_kernel(x: torch.Tensor) -> torch.Tensor:
    out = torch.empty_like(x)
    for tile in hl.tile(x.shape[0]):
        out[tile] = x[tile] * 2
    return out

x = torch.randn(100, device='cuda')
result = test_kernel(x)
torch.testing.assert_close(result, x * 2)
print("Verification successful!")

Development Dependencies

If you installed with [dev], you get additional development tools:

pytest - Test runner
pre-commit - Code formatting and linting hooks

Set up pre-commit hooks for development:

pre-commit install

Optional Dependencies

Documentation Building

To build documentation locally:

pip install -e '.[docs]'
cd docs
make html

GPU Requirements

Matches the requirements of Triton. At the time of writing:

NVIDIA GPUs (Compute Capability 8.0+)
AMD GPUs (ROCm 6.2+)
Via third-party forks: Intel XPU, CPU, and many other architectures

Next Steps

Once installation is complete:

Check out the API Reference for complete API documentation
Explore the examples/ folder for real-world patterns