CUDA Programming

A collection of CUDA programs demonstrating GPU computing techniques, primarily focused on matrix multiplication algorithms.

Files

• cuda_device_info.cu - CUDA device information and capabilities
• matrix_multiplication_basic.cu - Basic GPU matrix multiplication
• matrix_multiplication_benchmark.cu - Performance benchmarking with different block sizes
• matrix_multiplication_optimized.cu - Optimized matrix multiplication with timing
• matrix_multiplication_performance.cu - Performance-focused implementation
• tiled_matrix_multiplication.cu - Tiled algorithm using shared memory
• tiled_matrix_multiplication_advanced.cu - Advanced tiled implementation

Compilation

nvcc -o program_name program_name.cu

Features

• GPU vs CPU performance comparison
• Multiple matrix sizes (100x100 to 1500x1500)
• Different thread block configurations
• Shared memory optimization
• Memory transfer timing
• Result validation