GPU Programming with OpenACC Training Course

Introduction

• Defining OpenACC
• Comparing OpenACC with OpenCL, CUDA, and SYCL
• Overview of OpenACC features and architectural components
• Configuring the development environment

Getting Started

• Creating an OpenACC project within Visual Studio Code
• Exploring project structure and file organization
• Compiling and executing the program
• Displaying output using printf and fprintf

OpenACC Directives and Clauses

• Understanding OpenACC directives and clause usage
• Applying parallel directives to establish parallel regions
• Utilizing kernels directives for compiler-managed parallelism
• Using loop directives to parallelize iterations
• Managing data movement via data directives
• Synchronizing data with update directives
• Enhancing data reuse through cache directives
• Creating device functions using routine directives
• Synchronizing events with wait directives

OpenACC API

• Understanding the role of the OpenACC API
• Querying device information and capabilities
• Setting device number and type
• Handling errors and exceptions
• Creating and synchronizing events

OpenACC Libraries and Interoperability

• Understanding OpenACC libraries and interoperability standards
• Utilizing math, random, and complex number libraries
• Integrating with other programming models (CUDA, OpenMP, MPI)
• Integrating with specific GPU libraries (cuBLAS, cuFFT)

OpenACC Tools

• Understanding the role of OpenACC tools in the development workflow
• Profiling and debugging OpenACC programs
• Conducting performance analysis using PGI Compiler, NVIDIA Nsight Systems, and Allinea Forge

Optimization

• Identifying factors that impact OpenACC program performance
• Optimizing data locality and minimizing transfer overhead
• Enhancing loop parallelism and applying fusion techniques
• Optimizing kernel parallelism and fusion
• Improving vectorization and utilizing auto-tuning

Summary and Next Steps