This is a CUDA tutorial and question bank prepared specifically for beginners in high-performance computing (HPC), including 200 CUDA-implemented operators, study notes, and hands-on practice of self-written performance benchmarking against official HGEMM and FlashAttention-2. It is suitable for interview preparation related to model inference optimization and operator optimization.