Re: problem with gsl and icc

[Brian Gough <bjg at network-theory dot co dot uk>]

Daniel Rohe writes:
 > ok, so I compiled gsl using icc by setting the CC environment variable 
 > to icc using the tcsh.
 > compilation is not the problem, but make check yields the following:

 > FAIL: gsl_isinf(inf) (0 observed vs 1 expected)
 > FAIL: gsl_isinf(-inf) (0 observed vs -1 expected)
 > FAIL: gsl_isnan(nan) (0 observed vs 1 expected)
 > FAIL: gsl_finite(inf) (1 observed vs 0 expected)
 > FAIL: gsl_finite(nan) (1 observed vs 0 expected)

in sys/test.c find the part which says

  {
    double zero, one, inf, nan;
    int s;

    zero = 0.0;
    one = 1.0;
    inf = exp(1.0e10);
    nan = inf / inf;

and add print statements for the variables inf and nan, or look at them
in the debugger, to see if they are correct.

If they ok are then the problem is in sys/infnan.c where there are
tests like (x != x) which check for nans.  These tests require proper
IEEE arithmetic comparisons.  Some compilers turn off IEEE arithmetic
comparisons by default and need a special flag to turn it back on.

-- 
Brian

----------------------------------------------------------------------
Network Theory Ltd            Phone: 0117 3179309 (+44 117 3179309)
15 Royal Park                 WWW: http://www.network-theory.co.uk/
Bristol BS8 3AL               Email: bjg@network-theory.co.uk     
United Kingdom                
----------------------------------------------------------------------