4 comments

  • AlotOfReading23 minutes ago
    Usually, if you know enough about your algorithms to select an appropriate float alternative, you also know enough to fix your float code and that&#x27;s what you should actually do.<p>That said, some of these aren&#x27;t alternatives. Symbolic computation is a different thing entirely. Interval arithmetic can be built atop floats (e.g. IEEE-1788) and has its own zoo of unintuitive behaviors. BCD is better called a historical artifact than an alternative these days.<p>It&#x27;s really just rationals and decimal floats in this list, which probably don&#x27;t solve the issues you have if you&#x27;re considering float alternatives.
  • klodolph16 minutes ago
    Worth mentioning is storing the logarithm of the number.<p>Seen it used in a couple places. Logarithmic depth buffer is one. Yamaha DX7 is another.
    • mtklein5 minutes ago
      That&#x27;s kind of exactly what floats are. You store the log2 of the number, a bit for its sign, and in what remaining bits you have left some fixed-point scaling between adjacent powers.
  • schiffern1 day ago
    None of the base 10 formats, but the hobby calculation language Frink supports exact rational fractions, arbitrary width bigints (not pictured), intervals arithmetic, and symbolic expressions.<p><a href="https:&#x2F;&#x2F;frinklang.org&#x2F;fsp&#x2F;frink.fsp?fromVal=new+interval%5B-10%2C+1%2F5%2C+1%5D%5E2&amp;toVal=#calc" rel="nofollow">https:&#x2F;&#x2F;frinklang.org&#x2F;fsp&#x2F;frink.fsp?fromVal=new+interval%5B-...</a>
  • MiroslavPokorny2 days ago
    Strange the most obvious - fixed point numbers is not mentioned.
    • vismit20001 day ago
      <a href="https:&#x2F;&#x2F;wizardzines.com&#x2F;comics&#x2F;fixed-point&#x2F;" rel="nofollow">https:&#x2F;&#x2F;wizardzines.com&#x2F;comics&#x2F;fixed-point&#x2F;</a>
    • cryptonector25 minutes ago
      Nor unums!!