Merge pull request #378 from boostorg/cuda_float128_fix

Tentative fix for CUDA/Clang/Cray issue reported here https://github.…
This commit is contained in:
jzmaddock
2021-04-06 10:03:58 +01:00
committed by GitHub

View File

@ -475,6 +475,13 @@ namespace std {
# define BOOST_CTOR_TYPENAME # define BOOST_CTOR_TYPENAME
#endif #endif
//
// If we're on a CUDA device (note DEVICE not HOST, irrespective of compiler) then disable __float128 support if present:
//
#if defined(__CUDA_ARCH__) && defined(BOOST_HAS_FLOAT128)
# undef BOOST_HAS_FLOAT128
#endif
// long long workaround ------------------------------------------// // long long workaround ------------------------------------------//
// On gcc (and maybe other compilers?) long long is alway supported // On gcc (and maybe other compilers?) long long is alway supported
// but it's use may generate either warnings (with -ansi), or errors // but it's use may generate either warnings (with -ansi), or errors