PeterJensen

@simhos01 Not 100% sure I understand your suggestion. Do you suggest introducing SIMD.XIntAxB name spaces to hang the sign-agnostic operations on. It would be OK to use both IntAxB and...

Thanks @littledan That comment should've been made here. My mistake.

@simhos01 's comment above regarding massive use of `vreinterpretq_xnn_ymm()` in example code from [skia](https://github.com/google/skia/blob/master/src/opts/SkXfermode_opts_arm_neon.cpp), [webp](https://github.com/webmproject/libwebp/blob/master/src/dsp/dec_neon.c), and [vpx](https://github.com/webmproject/libvpx/blob/master/vp8/common/arm/neon/vp8_subpixelvariance_neon.c) where vreinterpret calls are used for type casts is pretty important. It provides...

@jfbastien Yes the conversions that change the number of elements don't count. My point was that the sign conversions do appear in real code, as witnessed by the _s8_u8 and...