Use _Float16 for half conversions if available#231
Use _Float16 for half conversions if available#231thiagomacieira merged 1 commit intointel:mainfrom kalcutter:float16-half-conversions
Conversation
thiagomacieira
left a comment
thiagomacieira
left a comment
There was a problem hiding this comment.
looks good, just the minor suggestion.
Recent ARM compilers support _Float16.
The reason for that is that x86 finally got it too (GCC 12 added support) and that C++ is getting this in P1467
I didn't mean that "recent". 😄 Even 3-year old clang supports this on ARM: https://godbolt.org/z/o3T5oWv57 |
Nice... clang 15 also has x86 support now finally. |
I didn't think |
Implement support for half-precision floating-point conversions using `_Float16` introduced in C11 extension ISO/IEC TS 18661-3.
2 participants
Implement support for half-precision floating-point conversions using
_Float16introduced in C11 extension ISO/IEC TS 18661-3.Recent ARM compilers support _Float16.