Re: Sin & Cos speed worries

"Everett M. Greene" wrote:

"Fred J. Tydeman" writes:

You might be surprised at how inaccurate many of the math libraries
are that ship with compilers. A "C" program to test sin(355.) and the
results can be found in tsin.c on the web page in my signature.
I would be interested in results people get that are not already listed.

sin(355) can be closely approximated by zero. What's
the problem?

He's worried about "closeness" of the C compilers. Comparing with
the leading Fortran compiler (CVF) in dble against his posted (quad?)
"correct" result shows they differ by approximately e-19 - pass!

--
sdx
http://www.sdynamix.com

.

Relevant Pages

  • Re: Debug optimized code
    ... with one set of optimizations and then ship untested/undebugged code ... compiled using a different set of optimizations. ... with a few exceptions that compilers dealt with by context. ... a developer for some implementation purpose. ...
    (comp.compilers)
  • Re: No more free non-security patches == No more Mozilla, Firefox, KDE etc.
    ... >> For the compilers themselves, ... >> lists which parts of the compilers you may ship with your ... of your application to that specific set of OS libraries ... clearly prefer to have your users be able to install ...
    (comp.unix.solaris)
  • Re: error invoking wish85.exe, tclsh85.exe
    ... You switched compilers. ... If you compiled 8.4.13 with VC6 + Platform SDK you don't need to ship ... I added the msvcrt80.dll in the same directory where these exe are, ...
    (comp.lang.tcl)
  • Re: erf function in C
    ... > Are there any C compilers which have the erf function (from ... > probability) as part of their math libraries? ...
    (comp.lang.c)
  • Re: erf function in C
    ... > Are there any C compilers which have the erf function (from ... > probability) as part of their math libraries? ... I'm aware of two compilers, one C and one Fortran (Watcom). ...
    (comp.lang.c)