llamacpp-installer

Llama.cpp Automated Installer

English

Automatic installer for llama.cpp with hardware-specific optimizations. The project detects your hardware type (Raspberry Pi 5, Raspberry Pi 4, Termux Android, Linux x86_64) and automatically selects optimal compilation flags for maximum performance.

Author: Fibogacci (https://fibogacci.pl)
License: MIT
Technologies: Python, Textual (GUI), Typer (CLI), Rich, psutil

Features

🔍 Automatic hardware detection: Raspberry Pi, Android Termux, Linux x86_64
⚡ Optimized compilation: Hardware-specific CMAKE flags for maximum performance
🖥️ Dual interface: Interactive GUI (Textual) and CLI (Typer)
🌍 Multi-language: Polish and English interface
📊 Real-time progress: Installation progress with detailed logging
🛠️ Custom configs: Support for user-defined optimization files
🔄 Dynamic detection: Automatic CPU capability testing (NEW)

Supported Hardware

Raspberry Pi

Raspberry Pi 5 (8GB/16GB) - Maximum ARM64 Cortex-A76 optimizations with OpenBLAS and RPC
Raspberry Pi 5 (4GB) - Balanced ARM64 optimizations with OpenBLAS
Raspberry Pi 4 - Cortex-A72 optimizations with OpenBLAS

Linux x86_64

Dynamic ⭐ - Automatic detection and optimization (RECOMMENDED)

⚠️ Warning: dynamic works only on x86_64. For ARM/Raspberry Pi use dedicated configurations (rpi5_8gb, rpi5_16gb, rpi4).
Standard - Full AVX2 optimizations with OpenBLAS (newer CPUs)
Legacy - AVX optimizations without AVX2 (2010-2013 CPUs)
Minimal - Basic optimizations (very old hardware)
No optimization - Widest compatibility (works everywhere)

Mobile

Termux Android - Minimal optimizations without BLAS

Quick Start

Installation

Direct Installation (Recommended)

# Clone repository
git clone https://github.com/fibogacci/llamacpp-installer.git
cd llamacpp-installer

# Install dependencies globally
pip install -r requirements.txt

Alternative: Virtual Environment (Optional)

If you have issues with global installation or want to isolate dependencies:

# Clone repository
git clone https://github.com/fibogacci/llamacpp-installer.git
cd llamacpp-installer

# Create virtual environment
python -m venv venv-llamacpp-installer

# Activate virtual environment
source venv-llamacpp-installer/bin/activate

# Install dependencies in virtual environment
pip install -r requirements.txt

# To deactivate virtual environment (after finishing work)
deactivate

Usage

GUI Interface (Recommended)

# Polish (default)
python main.py

# English
python main.py --lang en

CLI Interface

# Hardware detection
python cli.py detect
python cli.py detect --lang en

# Automatic installation (recommended)
python cli.py install --hardware dynamic --dir /path/to/install

# Manual hardware selection
python cli.py install --hardware rpi5_8gb --dir /home/user/llama
python cli.py install --hardware x86_linux --dir /opt/llama --lang en

# Custom configuration
python cli.py install --config example_configs/x86_avx512.txt --dir /path/to/install

Hardware Types

Hardware	Type	Description
Raspberry Pi 5 8/16GB	`rpi5_8gb`	Maximum performance with RPC
Raspberry Pi 5 4GB	`rpi5_4gb`	Balanced optimization
Raspberry Pi 4	`rpi4`	Cortex-A72 optimized
Linux x86_64 Auto	`dynamic`	Automatic detection ⭐
Linux x86_64 New	`x86_linux`	AVX2 + OpenBLAS
Linux x86_64 Legacy	`x86_linux_old`	AVX without AVX2
Linux x86_64 Minimal	`x86_linux_minimal`	Basic optimizations
No optimization	`no_optimization`	Maximum compatibility
Termux Android	`termux`	Mobile-optimized

Custom Configurations

Create your own .txt files with CMAKE flags:

# example_configs/my_config.txt
-DGGML_NATIVE=ON
-DGGML_AVX2=ON  
-DGGML_OPENMP=ON
-DGGML_OPENBLAS=ON

Usage:

python cli.py install --config my_config.txt --dir /path/to/install

Directory Structure

After installation, the structure will be:

/your/chosen/path/
└── llama.cpp/
    ├── build/
    │   └── bin/
    │       ├── llama-cli
    │       ├── llama-server  
    │       └── ...
    ├── logs/
    │   └── llamacpp_installer_*.log
    ├── llama-cli.sh      # Wrapper script
    ├── llama-server.sh   # Wrapper script
    └── llama-simple.sh   # Wrapper script

Language Support

CLI

# Polish (default)
python cli.py detect
python cli.py install --hardware dynamic --dir /path

# English  
python cli.py detect --lang en
python cli.py install --hardware dynamic --dir /path --lang en

GUI

# Polish (default)
python main.py

# English
python main.py --lang en
python main.py --en

Environment Variable

export LLAMACPP_INSTALLER_LANG=en
python main.py  # Will use English

Advanced Usage

Debug Mode

# Enable detailed logging
python cli.py detect --debug
python cli.py install --hardware dynamic --dir /path --debug

Testing Hardware Detection

# Test hardware detection
python hardware_detector.py

# Test dynamic CPU detection  
python dynamic_config.py

# Test optimization configs
python optimization_configs.py

Troubleshooting

Common Issues

Compilation errors on older CPUs
- Use --hardware x86_linux_old or --hardware dynamic
- Check logs in {install_dir}/logs/
Missing dependencies
- Install build essentials: sudo apt install build-essential cmake git
- For Ubuntu/Debian with OpenBLAS: sudo apt install libopenblas-dev
Permission errors
- Ensure write permissions to installation directory
- Use sudo only if installing to system directories
Termux specific issues
- Install required packages: pkg install python cmake git
- Use --hardware termux for mobile optimization

Log Files

Installation logs are saved in {installation_directory}/logs/ with format: llamacpp_installer_YYYYMMDD_HHMMSS.log

Requirements

Python 3.7+
Build tools (gcc, cmake, git)
Internet connection for downloading llama.cpp
Sufficient disk space (2-3 GB for full compilation)

Contributing

Fork the repository
Create feature branch
Make changes
Test on different hardware if possible
Submit pull request

License

MIT License - see LICENSE file for details.

Author

Fibogacci

Website: https://fibogacci.pl
GitHub: https://github.com/fibogacci

For Polish documentation, see README_PL.md

This site is open source. Improve this page.