AMD ROCm Debugging Overview
The TotalView debugger is an integrated debugging tool capable of simultaneously debugging HIP (Heterogeneous Interface for Portability) code running on both the host system and in the ROCm environment on an AMD GPU.
Supported major features:

Debug HIP applications running directly on AMD GPU hardware

Set breakpoints, pause execution, and single step in HIP code

Access runtime variables, such as threadIdx, blockIdx, blockDim, etc.

Debug multiple GPU devices per process

Debug remote, distributed, and clustered systems

Support for host debugging features
Requirements:
The AMD ROCm platform and a host distribution supported by AMD. For versions and supported AMD GPUs, see the
TotalView Supported Platforms Guide.