On Mon, Jul 24, 2023 at 07:39:05PM +0000, Joseph Myers wrote:
> On Mon, 24 Jul 2023, Jakub Jelinek via Gcc-patches wrote:
>
> > I believe it is only +- which has this problematic behavior and I think
>
> fma has the same property (of rounding-mode-dependent exact results), but
> I think that's not relevant here?
Indeed, real_arithmetics doesn't handle FMA* and I think the ranger doesn't
either as of now, if it would, it would likely use mpfr for that and indeed
would need to take this into account.
Jakub
