The source project of this merge request has been removed.
util: use msvcrt _fpreset
mingw initializes the fpu to higher precision than the standard 53 bits mantissa
this is needed to fix convert_float_to_8unorm and pass the u_format_test test in 32 bits mode
this source is taken from mingw-w64's crt and is public domain
https://github.com/mirror/mingw-w64/blob/master/mingw-w64-crt/crt/CRT_fp8.c
Edited by xantares