Learn GPU execution concepts and the AMD ROCm™ software stack, and gain hands-on experience in HIP (Heterogenous-computing Interface for Portability) programming. Set up development environments, build and debug HIP kernels, migrate CUDA applications using hipify, apply optimization strategies, and complete an end-to-end GPU migration case study.
This course covers:
- Understanding GPU execution fundamentals and AMD GPU architecture concepts
- Exploring the AMD ROCm™ platform components and toolchain
- Developing HIP kernels and understanding host–device execution workflows
- Mapping CUDA concepts, APIs, and memory models to HIP equivalents
- Using hipify tools to automate CUDA-to-HIP
- Debugging, profiling, and optimizing applications on AMD GPUs
- Applying migration workflows through an end-to-end CUDA-to-ROCm™ case study
