Fix HIP debug flags to use -G only for NVIDIA compilers

lattice / quda

QUDA is a library for performing calculations in lattice QCD on GPUs.

https://lattice.github.io/quda

Other

294 stars 100 forks source link

Closed jakurzak closed 1 day ago

jakurzak commented 4 days ago

Updated CMake configuration to set -g -G only when the HIP compiler is NVIDIA (nvcc).
For Clang-based HIP compilers, use -g alone as it supports both host and device debugging.
Improved flag descriptions in CMake cache to reflect the specific behavior of each compiler.
This ensures compatibility and avoids unsupported flag errors with Clang.