Fundamentals of Accelerated Computing with CUDA C/C++ Training Course

GPU Computing and CUDA Architecture

• Differences between CPU and GPU architectures
• NVIDIA GPU streaming multiprocessor architecture
• Overview of the CUDA programming model
• Heterogeneous computing and the host-device paradigm

Establishing the CUDA Development Environment

• Installation of CUDA Toolkit 13.x
• NVCC compiler and build processes
• Verifying the environment through device queries
• IDE integration and development tools

Writing and Launching CUDA Kernels

• Syntax and qualifiers for kernel functions
• Launch configuration and execution processes
• Vector addition and fundamental data-parallel patterns
• CUDA error-checking macros

CUDA Thread Hierarchy and Execution Model

• Organization of grids, blocks, and threads
• Thread indexing and global ID calculation
• Warp execution and the SIMT model
• Occupancy and resource utilization metrics

GPU Memory Architecture and Management

• Memory types: global, shared, constant, and registers
• Allocating and deallocating device memory
• Data transfers between host and device
• Using shared memory for intra-block collaboration

Unified Memory and Data Migration

• The unified memory model and managed allocations
• Page migration and on-demand paging mechanisms
• Asynchronous prefetching using cudaMemPrefetchAsync
• Memory advice hints for optimizing access patterns

System-Wide Profiling with Nsight Systems

• Timeline analysis in Nsight Systems
• Identifying CPU-GPU synchronization points
• Visualizing kernel execution and memory transfers
• Interpreting system-level performance data

Kernel Optimization with Nsight Compute

• Interactive kernel profiling with Nsight Compute
• Analyzing memory throughput and bandwidth
• Evaluating compute utilization and warp state statistics
• Guided analysis and optimization strategies

Concurrent Streams and Asynchronous Operations

• CUDA streams and the default stream behavior
• Overlapping kernel execution with data transfers
• Stream synchronization and CUDA events
• Design patterns for multi-stream pipelines

Error Handling and Debugging Tools

• CUDA API error codes and recovery strategies
• Using compute-sanitizer for memory access validation
• Kernelf debugging with cuda-gdb
• Assertions and synchronous error detection methods

Profile-Driven Optimization Workflow

• Iterative profiling methodology
• Identifying and prioritizing bottlenecks
• Performance regression testing
• Documenting optimization decisions

End-to-End Accelerated Application Project

• Designing a comprehensive GPU-accelerated solution
• Integrating profiling throughout the development lifecycle
• Performance benchmarking and reporting
• Deployment considerations for production environments