Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.arch    |    Apparently more than just beeps & boops    |    131,241 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 130,377 of 131,241    |
|    John Savard to Thomas Koenig    |
|    Re: Why no roundTiesToOdd in IEEE?    |
|    28 Nov 25 05:19:31    |
      From: quadibloc@invalid.invalid              On Mon, 10 Nov 2025 06:59:19 +0000, Thomas Koenig wrote:              > There is a clear use case for roundTiesToOdd (correctly rounding       > calculations for later conversion to a lower precision), this is why       > Power implements it for its 128-bit HW.              I see that what IEEE 754 does have is round to nearest with ties to even;       allegedly, according to the Wikipedia article, this is the recommended       option. There is also a round to nearest with ties to "away", that is, to       the larger magnitude; this is provided as an option for decimal arithmetic       only.              If rounding ties to odd makes conversion to a lower precision more likely       to be correct, I'm surprised that it isn't the preferred option. But I       think I'd have to learn more about numerical analysis to make an       intelligent comment on this matter.              John Savard              --- SoupGate-Win32 v1.05        * Origin: you cannot sedate... all the things you hate (1:229/2)    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca