first commit

Author: stephen2ml

README.md

@@ -0,0 +1,350 @@
+# GPU Programming 101 🚀
+
+A comprehensive hands-on course for learning GPU programming with CUDA and HIP, covering fundamental concepts through advanced optimization techniques.
+
+## 🎯 Course Overview
+
+This course provides practical, hands-on experience with GPU programming, covering everything from basic parallel computing concepts to advanced optimization techniques. Each module contains theory, working code examples, and exercises.
+
+## 📚 Course Structure
+
+### Module 1: Foundations of GPU Computing ✅
+**Status**: Complete  
+**Duration**: 4-6 hours  
+**Level**: Beginner  
+
+**Topics Covered**:
+- GPU architecture and SIMT execution model
+- CUDA and HIP programming fundamentals
+- Memory management and data transfers
+- Basic parallel execution patterns
+- Debugging and optimization basics
+
+**[📁 Go to Module 1](modules/module1/)**
+
+### Module 2: Multi-Dimensional Data Processing ✅
+**Status**: Complete  
+**Duration**: 6-8 hours  
+**Level**: Beginner-Intermediate  
+
+**Topics Covered**:
+- Multidimensional grid organization
+- Thread mapping to data structures
+- Image processing kernels
+- Matrix multiplication algorithms
+- Advanced memory management
+
+**[📁 Go to Module 2](modules/module2/)**
+
+### Module 3: GPU Architecture and Execution Models ✅
+**Status**: Complete  
+**Duration**: 6-8 hours  
+**Level**: Intermediate  
+
+**Topics Covered**:
+- GPU architecture deep dive
+- Warp scheduling and SIMD hardware
+- Control divergence and optimization
+- Resource partitioning and occupancy
+- Advanced parallel patterns
+
+**[📁 Go to Module 3](modules/module3/)**
+
+### Module 4: Advanced GPU Programming Techniques ✅
+**Status**: Complete  
+**Duration**: 8-10 hours  
+**Level**: Intermediate-Advanced  
+
+**Topics Covered**:
+- Multi-GPU programming and scalability
+- Asynchronous execution with streams
+- Dynamic parallelism techniques
+- Advanced memory optimization
+- Cross-platform development strategies
+
+**[📁 Go to Module 4](modules/module4/)**
+
+### Module 5: Performance Engineering and Optimization ✅
+**Status**: Complete  
+**Duration**: 6-8 hours  
+**Level**: Advanced  
+
+**Topics Covered**:
+- Performance profiling and analysis
+- Memory bandwidth optimization
+- Kernel optimization strategies
+- Bottleneck identification and resolution
+- Production performance engineering
+
+**[📁 Go to Module 5](modules/module5/)**
+
+### Module 6: Fundamental Parallel Algorithms 🚧
+**Status**: Planned  
+**Duration**: 8-10 hours  
+**Level**: Intermediate-Advanced  
+
+**Topics**:
+- Convolution and filtering algorithms
+- Stencil computations
+- Histogram and atomic operations
+- Reduction patterns and optimizations
+- Prefix sum (scan) algorithms
+
+### Module 7: Advanced Algorithmic Patterns 🚧
+**Status**: Planned  
+**Duration**: 8-10 hours  
+**Level**: Advanced  
+
+**Topics**:
+- Merge and sorting algorithms
+- Sparse matrix computations
+- Graph traversal algorithms
+- Dynamic programming on GPU
+- Load balancing techniques
+
+### Module 8: Domain-Specific Applications 🚧
+**Status**: Planned  
+**Duration**: 10-12 hours  
+**Level**: Advanced  
+
+**Topics**:
+- Deep learning inference kernels
+- Scientific computing applications
+- Image and signal processing
+- Monte Carlo simulations
+- Numerical methods optimization
+
+### Module 9: Production GPU Programming 🚧
+**Status**: Planned  
+**Duration**: 6-8 hours  
+**Level**: Expert  
+
+**Topics**:
+- Cluster computing with MPI
+- Dynamic parallelism patterns
+- Performance regression testing
+- Cross-platform deployment
+- Future GPU architectures
+
+## 🛠️ Prerequisites
+
+### Hardware Requirements
+- **NVIDIA GPU**: GeForce GTX 1060 or better, or Tesla/Quadro equivalent
+- **OR AMD GPU**: RX 580 or better, with ROCm support
+- **Memory**: 8GB+ system RAM, 4GB+ GPU memory recommended
+
+### Software Requirements
+- **Operating System**: Linux (recommended), Windows 10/11, or macOS
+- **CUDA Toolkit**: 11.0+ for NVIDIA GPUs
+- **ROCm**: 4.0+ for AMD GPUs
+- **Compiler**: GCC 7+, Clang 8+, or MSVC 2019+
+- **Build Tools**: Make, CMake (optional)
+
+### Programming Knowledge
+- **C/C++**: Intermediate level (pointers, memory management, basic OOP)
+- **Command Line**: Basic terminal/shell usage
+- **Math**: Linear algebra basics helpful but not required
+
+## 🚀 Quick Start
+
+### Option 1: Docker (Recommended)
+Perfect for getting started without installing CUDA/ROCm on your host system.
+
+```bash
+git clone https://github.com/yourusername/gpu-programming-101.git
+cd gpu-programming-101
+
+# Test your Docker setup
+./docker/scripts/test.sh
+
+# Build development container  
+./docker/scripts/build.sh --all
+
+# Auto-detect GPU and start appropriate container
+./docker/scripts/run.sh --auto
+
+# Inside container - test your GPU
+/workspace/test-gpu.sh
+
+# Start learning!
+cd modules/module1 && cat README.md
+```
+
+### Option 2: Native Installation
+
+```bash
+git clone https://github.com/yourusername/gpu-programming-101.git
+cd gpu-programming-101
+
+# Check system requirements
+# For NVIDIA systems
+nvidia-smi && nvcc --version
+
+# For AMD systems  
+rocm-smi && hipcc --version
+
+# Start with Module 1
+cd modules/module1
+cat README.md  # Read module overview
+cd examples
+make           # Build examples
+./04_device_info_cuda  # Check your GPU
+```
+
+### 🐳 Docker Benefits
+- **No host setup required**: Complete development environment in containers
+- **Multi-platform**: Test both CUDA and HIP code easily  
+- **Consistent environment**: Same setup across different systems
+- **Integrated tools**: Profilers, debuggers, and Jupyter Lab included
+- **Easy cleanup**: Remove containers when done
+
+**[📖 Full Docker Guide](docker/README.md)**
+
+### 4. Follow Learning Path
+Each module contains:
+- **README.md** - Module overview and learning objectives
+- **content.md** - Comprehensive theory and explanations  
+- **examples/** - Working code examples with build system
+- **exercises/** - Additional practice problems (when available)
+
+## 📁 Project Structure
+
+```
+gpu-programming-101/
+├── README.md                    # This file
+├── SUMMARY.md                   # Detailed curriculum overview
+├── Makefile                     # Project-wide build system
+└── modules/
+    ├── module1/                 # Heterogeneous Data Parallel Computing
+    │   ├── README.md           # Module overview
+    │   ├── content.md          # Theory and explanations
+    │   └── examples/           # Working code examples
+    │       ├── Makefile
+    │       ├── README.md
+    │       └── *.cu, *.cpp     # Source files
+    ├── module2/                 # Multidimensional Grids and Data
+    │   └── [Coming Soon]
+    ├── module3/                 # Compute Architecture and Scheduling  
+    │   └── [Coming Soon]
+    └── [Additional Modules]
+```
+
+## 🎓 Learning Path Recommendations
+
+### For Complete Beginners
+1. **Start with Module 1** - Focus on understanding basic concepts
+2. **Practice extensively** - Modify examples and experiment
+3. **Use debugging tools** - Learn proper error handling
+4. **Progress gradually** - Master each concept before moving on
+
+### For Experienced Programmers
+1. **Skim Module 1 theory** - Focus on GPU-specific concepts
+2. **Run all examples** - Understand performance characteristics
+3. **Jump to specific topics** - Use course as reference material
+4. **Contribute improvements** - Help expand the course content
+
+### For Researchers/Scientists
+1. **Focus on relevant modules** - Skip graphics-specific content
+2. **Emphasize performance** - Pay special attention to optimization
+3. **Explore libraries** - Learn cuBLAS, cuFFT, Thrust, etc.
+4. **Real-world applications** - Adapt examples to your domain
+
+## 🔧 Build System
+
+### Project-wide Build
+```bash
+# Build all available modules
+make all
+
+# Build specific module
+make module1
+
+# Clean all builds
+make clean
+
+# Run tests
+make test
+```
+
+### Module-specific Build
+```bash
+cd modules/module1/examples
+make          # Build all examples
+make vector_add_cuda  # Build specific example
+make test     # Run module tests
+```
+
+## 🐛 Troubleshooting
+
+### Common Setup Issues
+
+**"nvcc: command not found"**
+```bash
+export PATH=/usr/local/cuda/bin:$PATH
+export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
+```
+
+**"No CUDA-capable device found"**
+- Check `nvidia-smi` shows your GPU
+- Verify driver installation
+- Ensure GPU is not in exclusive/prohibited mode
+
+**"HIP compilation failed"**
+```bash
+# For AMD GPUs
+export HIP_PLATFORM=amd
+
+# For NVIDIA GPUs with HIP
+export HIP_PLATFORM=nvidia
+```
+
+### Getting Help
+- **Module Issues**: Check module-specific README files
+- **Code Problems**: Look at debugging examples in Module 1
+- **Performance**: Use profiling tools covered in later modules
+- **Community**: Create an issue in the repository for help
+
+## 📖 Additional Resources
+
+### Official Documentation
+- [CUDA Programming Guide](https://docs.nvidia.com/cuda/cuda-c-programming-guide/)
+- [HIP Programming Guide](https://rocmdocs.amd.com/en/latest/Programming_Guides/HIP-GUIDE.html)
+- [ROCm Documentation](https://rocmdocs.amd.com/)
+
+### Books
+- "CUDA by Example" - Sanders, Kandrot
+- "Professional CUDA C Programming" - Cheng, Grossman, McKercher
+- "GPU Computing Gems" - NVIDIA Corporation
+
+### Online Resources
+- [NVIDIA Developer Zone](https://developer.nvidia.com/)
+- [AMD Developer Central](https://developer.amd.com/)
+- [GPU Computing Community](https://forums.developer.nvidia.com/)
+
+## 🤝 Contributing
+
+We welcome contributions! Please follow standard open source contribution practices.
+
+### Ways to Contribute
+- **Add examples** for existing modules
+- **Create new modules** following the established structure
+- **Improve documentation** and fix typos
+- **Add exercises** and solutions
+- **Port examples** between CUDA and HIP
+- **Performance optimizations** and benchmarks
+
+## 📝 License
+
+This course is released under an open source license. Feel free to use, modify, and distribute for educational purposes.
+
+## 🏆 Acknowledgments
+
+- Thanks to the CUDA and ROCm development communities
+- Inspired by hands-on learning approaches in parallel computing education
+- Built with contributions from GPU programming educators and practitioners
+
+---
+
+**Happy GPU Programming!** 🚀⚡️
+
+*Last Updated: September 2025*