roofline-radar

Roofline analysis of AI models

App Features

• Interactive sidebar with configuration options
• Line, bar, and scatter charts using Plotly
• Data summary statistics
• Pie chart for category distribution
• Key metrics display
• Expandable data explorer
• User input demo

Running Locally

Using docker-compose (recommended)

docker-compose up --build

Using docker directly

docker build -t roofline-radar .
docker run -p 8501:8501 roofline-radar

Then open http://localhost:8501 in your browser.

Deploy to Heroku

Prerequisites

• Heroku CLI installed
• A Heroku account

Step 1: Create Heroku-specific files

Create a setup.sh file:

mkdir -p ~/.streamlit/

echo "\
[server]\n\
headless = true\n\
port = \$PORT\n\
enableCORS = false\n\
\n\
" > ~/.streamlit/config.toml

Create a Procfile:

web: sh setup.sh && streamlit run app.py

Step 2: Deploy

# Login to Heroku
heroku login

# Create a new Heroku app
heroku create roofline-radar

# Deploy using Git
git add .
git commit -m "Prepare for Heroku deployment"
git push heroku main

# Open the app
heroku open

Step 3: View logs (if needed)

heroku logs --tail

Alternative: Deploy with Docker on Heroku

# Login to Heroku container registry
heroku container:login

# Create a new app (if not already created)
heroku create roofline-radar

# Build and push the Docker image
heroku container:push web -a roofline-radar

# Release the container
heroku container:release web -a roofline-radar

# Open the app
heroku open -a roofline-radar

Note: For Docker deployment on Heroku, update the Dockerfile to use the $PORT environment variable:

ENTRYPOINT ["sh", "-c", "streamlit run app.py --server.port=$PORT --server.address=0.0.0.0"]