CUDA

CUDA is a general-purpose parallel computing platform and programming model developed by NVIDIA to leverage the massive parallel processing power of GPUs. CUDA's scalable programming model uses abstractions like thread group hierarchies, shared memories, and barrier synchronization to enable fine-grained and coarse-grained parallelism, allowing programs to automatically scale across various GPU architectures, by adjusting the number of multiprocessors and memory partitions.

CUDA C++ Programming

• vector_add.cu
• matrix_mul.cu
• reduction.cu
• stream_example.cu
• unified_memory.cu

Further Reading

• CUDA Toolkit Documentation
• CUDA C++ Programming Guide