[FFmpeg-devel] [PATCH v2 2/7] avutil/half2float: adjust conversion of NaN

Tomas Härdin tjoppen at acc.umu.se
Thu Aug 18 11:51:04 EEST 2022


sön 2022-08-14 klockan 18:48 +0200 skrev Timo Rothenpieler:
> IEEE-754 differentiates two different kind of NaNs.
> Quiet and Signaling ones. They are differentiated by the MSB of the
> mantissa.
> 
> For whatever reason, actual hardware conversion of half to single
> always
> sets the signaling bit to 1 if the mantissa is != 0, and to 0 if it's
> 0.
> So our code has to follow suite or fate-testing hardware float16 will
> be
> impossible.

Does IEEE-754 specify this behavior?

/Tomas



More information about the ffmpeg-devel mailing list