Comment by mathiasgredal

Having looked briefly at the code I still think C++17 parallel algorithms are more ergonomic compared to OpenMP: https://rocm.blogs.amd.com/software-tools-optimization/hipst...