Re: double precision g77

In article <pan.2006.08.22.17.08.17.913603@xxxxxxxxxxxxxxx>,
Richard Pitman <richard.pitman3@xxxxxxxxxxxxxxx> writes:

Many thanks for your responses, however this is still eluding me.

Since you're using g77, this is Fortran 77 (with some extensions).
Run your code through the ftnchek utility. Fix all warnings and
errors.

I am using IMPLICIT NONE at the start of all my routines and have done
a "grep -i real *.f" to check I haven't missed a declaration. All the
literal constants are in the form 2.0D0 and all functions are declared as
external.

Does the code use EQUIVALENCE?

Do I need to tell a subroutine that the arguments being passed to it are
double precision, other than by declaring them as such with a double
precision declaration in the calling routine and the subroutine?

If you declared the actual arguments and the corresponding
dummy arguments as double precision, then you should be set.

Have you checked that the algorithm is robust to the change in
precision? Are the any "if (x .eq. 3.14159)" type statements?
Comparison for equality of floating point values is tricky.

--
Steve
http://troutmask.apl.washington.edu/~kargl/
.

Relevant Pages

  • Re: IMPLICIT NONE (F2k8+/-)
    ... > I thought that you were using implicit typing. ... DOUBLE PRECISION FUNCTION. ... declaring INTEGER arrays in a type statement, ...
    (comp.lang.fortran)
  • Re: String data - right truncation error with numeric fields!!!
    ... When you are declaring your variables, ... precision and scale with your data types. ... data types with precision and scale - e.g. ...
    (microsoft.public.sqlserver.odbc)
  • Re: What is Negative Energy ???
    ... and then declaring, on the basis on his declaration to being a ... I 'll remind him of that standard applied to him, ... a lack of precision, to remind him a reasonable standard of precision ...
    (sci.physics.relativity)
  • Re: Any F2003 help translating C++ map syntax?
    ... > as a part of the language than MPI and BLAS. ... In fact, Fortran already ... I would like to see a way of declaring your own "attributes" ... no deficiencies and the other way is to make it so complicated ...
    (comp.lang.fortran)
  • Re: Pascal recommendation?
    ... Michael Vilain wrote: ... declaring something before using it is hardly upside down. ... FORTRAN could do it. ... A good FORTRAN programmer can write a FORTRAN program in *any* language. ...
    (comp.sys.mac.system)