From: monnier@iro.umontreal.ca   
      
   MitchAlsup [2026-01-11 18:18:00] wrote:   
   > Michael S  posted:   
   >> Terje Mathisen  wrote:   
   >>> Yeah, there are strong forces who want to have, at least as a   
   >>> suggested/recommended option, a set of transcendental functions which   
   >>> are exactly rounded.   
   >> I wonder who are those forces and what is the set they push for.   
      
   One reason to want it comes from portability and bit-for-bit   
   reproducibility.  These requirements don't actually care about the   
   rounding being *correct* as much as the rounding always being the same   
   across different hardware and libm implementations, but it seems rather   
   unlikely that the various actors involved would agree on a particular   
   return value if it's not the correctly-rounded one, so in practice this   
   becomes a push for correctly rounded results.   
      
   > The problem, here, is that even when one gets all the rounding correct,   
   > one has still lost various algebraic identities.   
   >   
   >      CRSIN(x)^2 + CRCOS(X)^2 ~= 1.0   
      
   Which properties are preserved and which ones aren't is inevitably   
   a compromise since, for example, the above one cannot be preserved   
   without breaking several others.   
      
      
   - Stefan   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)