cudair is an innovative tool designed to enable live-reloading for CUDA applications, streamlining your development process much like golang-air. Its straightforward configuration and Docker support make it an ideal choice for developers working with CUDA.

Key Features

• Live-Reloading: Instantly see the effects of your changes without the need for manual restarts, enhancing workflow efficiency.
• Docker Compatibility: Easily integrate cudair into your existing Docker environments for seamless development.
• Simple Configuration: Get started quickly with easy setup and minimal configuration.

Supported Go Version

For optimal performance, cudair recommends using Go version v1.23 or higher.

Quick Start Guide

To begin using cudair, navigate to your project directory and initiate cudair:

cd /path/to/your_projects
cudair run -c .cudair.toml

You can also initialize a project with a configuration file:

cudair init

Configuration Setup

Your configuration file must be in .toml format. Here’s a sample configuration:

# Root directory to watch
root = "."
# Temporary directory name
tmp_dir = "tmp"

[build]
# Executable command
bin = "./tmp/main"
# Compilation command
cmd = "nvcc --std=c++17 -o ./tmp/main main.cu"
# Log file for build errors
log = "build-errors.log"
# Directories to exclude from monitoring
exclude_dir = ["tmp"]

Using cudair with Docker

For developers utilizing Docker, cudair components are easily configured within Docker installations. Here’s a quick overview:

Requirements

• NVIDIA Driver
• NVIDIA Container Toolkit

Sample Dockerfile

FROM nvcr.io/nvidia/cuda:12.6.2-cudnn-devel-ubuntu20.04

RUN apt-get update && apt-get install -y wget

# Install Golang
RUN rm -rf /usr/local/go && wget https://go.dev/dl/go1.23.4.linux-amd64.tar.gz && tar -C /usr/local -xzf go1.23.4.linux-amd64.tar.gz

ENV PATH=$PATH:/usr/local/go/bin:/root/go/bin

RUN go install github.com/ei-sugimoto/cudair@latest

WORKDIR /app

COPY . .

CMD ["cudair", "run", "-c", ".cudair.toml"]

Building and Running Docker Container

To build your Docker image, execute:

docker build -t cudair-sample

Then, create the container with:

docker run --rm -it --gpus all -v path/to/your_project:/app cudair-sample

With cudair, you can significantly improve your CUDA application development experience, enhancing productivity and simplifying the workflow.