Skip to content

gui.md

Latest commit

 

History

History
73 lines (56 loc) · 2.2 KB

gui.md

File metadata and controls

73 lines (56 loc) · 2.2 KB

Interactive GUI Documentation

The GUI interface provides an interactive way to explore and understand GPU acceleration examples. This document explains how to use the interface and what each section demonstrates.

Running the GUI

  1. Install dependencies:

    pip install -r requirements.txt

  2. Launch the GUI:

    ./run_gui.sh

    Or run directly with Streamlit:

    streamlit run src/gui/app.py

Available Examples

1. Pandas Acceleration

  • Demonstrates how to accelerate pandas operations using cuDF
  • Interactive data generation with configurable size
  • Real-time performance comparison between CPU and GPU
  • Visualization of speedup achieved

2. UMAP & HDBSCAN

  • Shows acceleration of dimensionality reduction and clustering
  • Interactive parameter configuration
  • Side-by-side comparison of CPU and GPU implementations
  • Visual representation of the results

3. NetworkX GPU

  • Demonstrates graph analytics acceleration
  • Interactive graph generation
  • Performance comparison of common graph algorithms
  • Visual representation of speedup

Features

  • Interactive Controls: Adjust parameters and see results in real-time
  • Visual Comparisons: Clear visualizations of performance differences
  • Code Examples: View and copy example code for each acceleration technique
  • Performance Metrics: Real-time computation of speedup factors

Tips for Best Results

  1. Hardware Requirements:

    • NVIDIA GPU with CUDA support
    • CUDA Toolkit 11.0 or later
    • Sufficient GPU memory for larger datasets

  2. Performance Optimization:

    • Start with smaller datasets to understand the behavior
    • Gradually increase data size to see scaling effects
    • Monitor GPU memory usage for large operations

  3. Common Issues:

    • If you see "CUDA out of memory" errors, reduce the data size
    • For best performance, ensure no other GPU-intensive tasks are running
    • Some operations may require specific GPU architectures for optimal performance

Contributing

Feel free to contribute to the GUI by:

  1. Adding new examples
  2. Improving visualizations
  3. Optimizing performance
  4. Adding new features

Submit pull requests to the GitHub repository.