From: anton@mips.complang.tuwien.ac.at   
      
   Paul Rubin  writes:   
   >anton@mips.complang.tuwien.ac.at (Anton Ertl) writes:   
   >> So just use the same implementations of transcentental functions, and   
   >> your results will be bit-identical   
   >   
   >Same implementations = same FP operations in the exact same order?   
      
   Same operations with the same data flow.  Independent operations can   
   be reordered.   
      
   >That   
   >seems hard to ensure, if the functions are implemented in a language   
   >that leaves anything up to a compiler.   
      
   Even gcc heeds data flow of FP operations unless you tell it with   
   -fastmath that anything goes.   
      
   >Also, in the early implementations x87, 68881, NS320something(?),   
   >transcententals were included in the coprocessor and the workings   
   >weren't visible.   
      
   The bigger problem with at least x87 is that math you don't always get   
   bit-identical results even for basic operations such as addition,   
   thanks to double rounding.  So even if you implement transcendentals   
   yourself based basic operations, you can see results that are not   
   bit-identical.   
      
   - anton   
   --   
   M. Anton Ertl  http://www.complang.tuwien.ac.at/anton/home.html   
   comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html   
        New standard: https://forth-standard.org/   
   EuroForth 2023 proceedings: http://www.euroforth.org/ef23/papers/   
   EuroForth 2024 proceedings: http://www.euroforth.org/ef24/papers/   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)