llvm-project

ce73cbb6 - clang: Use generic builtins in cuda complex builtins header (#171106)

clang: Use generic builtins in cuda complex builtins header (#171106) There's no reason to use the ocml or nv prefixed functions and maintain this list of alias macros. I left these macros in for NVPTX in the scalbn and logb case, since those have a special case hack in the AMDGPU codegen and probably do not work on ptx.