NVIDIA/cuda-quantum

Refactor and remove the backendConfig auxiliary code

schweitzpgi opened this issue · 0 comments

At present the driver builds some extra code, backendConfig.cpp, to add information to the binary about what target is selected. I believe it does this to make the information available in both library and MLIR mode. As MLIR mode becomes the default (#1015 ) and library mode becomes deprecated (particularly with NVQC) this stub generation should be done by the compiler itself rather than via extra auxiliary sources.

See also #1574