AMD GPU Programming Training Course

Introduction

• Overview of ROCm
• Introduction to HIP
• Comparison: ROCm vs. CUDA vs. OpenCL
• Architecture and key features of ROCm and HIP
• Configuring the Development Environment

Getting Started

• Initializing a new ROCm project in Visual Studio Code
• Exploring project structure and file organization
• Compiling and executing the program
• Displaying output via printf and fprintf

ROCm API

• The role of the ROCm API within the host program
• Querying device information and capabilities using the ROCm API
• Allocating and freeing device memory via the ROCm API
• Transferring data between host and device using the ROCm API
• Launching kernels and synchronizing threads with the ROCm API
• Handling errors and exceptions using the ROCm API

HIP Language

• The role of the HIP language within the device program
• Writing GPU-executable kernels in HIP for data manipulation
• Utilizing HIP data types, qualifiers, operators, and expressions
• Employing HIP built-in functions, variables, and libraries for common operations

ROCm and HIP Memory Model

• Distinguishing between host and device memory models
• Working with ROCm and HIP memory spaces: global, shared, constant, and local
• Managing ROCm and HIP memory objects such as pointers, arrays, textures, and surfaces
• Applying ROCm and HIP memory access modes, including read-only, write-only, and read-write
• Understanding the ROCm and HIP memory consistency model and synchronization mechanisms

ROCm and HIP Execution Model

• Distinguishing between host and device execution models
• Defining parallelism using ROCm and HIP threads, blocks, and grids
• Utilizing ROCm and HIP thread functions, e.g., hipThreadIdx_x, hipBlockIdx_x, hipBlockDim_x
• Using ROCm and HIP block functions, e.g., __syncthreads, __threadfence_block
• Applying ROCm and HIP grid functions, e.g., hipGridDim_x, hipGridSync, and cooperative groups

Debugging

• Identifying common errors and bugs in ROCm and HIP programs
• Inspecting variables, breakpoints, and call stacks using the Visual Studio Code debugger
• Debugging ROCm and HIP programs on AMD devices with the ROCm Debugger
• Analyzing ROCm and HIP programs on AMD devices using the ROCm Profiler

Optimization

• Factors influencing the performance of ROCm and HIP programs
• Enhancing memory throughput with ROCm and HIP coalescing techniques
• Reducing memory latency via ROCm and HIP caching and prefetching methods
• Optimizing memory access and bandwidth using ROCm and HIP shared and local memory strategies
• Measuring and improving execution time and resource utilization with ROCm and HIP profiling tools

Summary and Next Steps