From: dxforth@gmail.com   
      
   On 14/07/2025 4:04 pm, Anton Ertl wrote:   
   > dxf  writes:   
   >> On 13/07/2025 7:01 pm, Anton Ertl wrote:   
   >>> ...   
   >>> For Forth, Inc. and MPE AFAIK their respective IA-32 Forth system was   
   >>> the only one with hardware FP for many years, so there probably was   
   >>> little pressure from users for bit-identical results with, say, SPARC,   
   >>> because they did not have a Forth system that ran on SPARC.   
   >>   
   >> What do you mean by "bit-identical results"?  Since SSE2 comes without   
   >> transcendentals (or basics such as FABS and FNEGATE) and implementers   
   >> are expected to supply their own, if anything, I expect results across   
   >> platforms and compilers to vary.   
   >   
   > There are operations for which IEEE 754 specifies the result to the   
   > last bit (except that AFAIK the representation of NaNs is not   
   > specified exactly), among them F+ F- F* F/ FSQRT, probably also   
   > FNEGATE and FABS.  It does not specify the exact result for   
   > transcendental functions, but if your implementation performs the same   
   > bit-exact operations for computing a transcendental function on two   
   > IEEE 754 compliant platforms, the result will be bit-identical (if it   
   > is a number).  So just use the same implementations of transcentental   
   > functions, and your results will be bit-identical; concerning the   
   > NaNs, if you find a difference, check if the involved values are NaNs.   
      
   So in mandating bit-identical results, not only in calculations but also   
   input/output, IEEE 754 is all about giving the illusion of truth in   
   floating-point when, if anything, they should be warning users don't be   
   fooled.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)