[RFC] Full float128, third iteration

Steve Kargl sgk@troutmask.apl.washington.edu
Fri Sep 10 01:15:00 GMT 2010


On Thu, Sep 09, 2010 at 10:58:10PM +0200, FX wrote:
> 
> Thanks to everyone who tested the two previous patches on their systems, I hope we're now close to the final integration (at least from the technical point of view!).
> 
> FX

The most basic of testing on x86_64-*-freebsd is showing
some promise

                === gfortran Summary ===

# of expected passes            36823
# of unexpected failures        37
# of unexpected successes       7
# of expected failures          59
# of unresolved testcases       37
# of unsupported tests          69
/usr/home/sgk/gcc/obj4x/gcc/testsuite/gfortran/../../gfortran  version 4.6.0 20100909 (experimental) (GCC) 

The 37 failures are do to lto and FreeBSD's libelf.

All of the numeric intrinsic functions give the expected
values.

troutmask:sgk[277] gfc4x -o inquiry inquiry.f90 && ./inquiry
       kind:    4     8     10     16
     digits:   24    53     64    113
  precision:    6    15     18     33
minexponent: -125 -1021 -16381 -16381
maxexponent:  128  1024  16384  16384
      range:   37   307   4931   4931
      radix:    2     2      2      2
tiny:
  single: 1.175494E-38
  double: 2.2250738585072014E-308
extended: 3.362103143112093506E-4932
    quad: 3.362103143112093506262677817321753E-4932
huge:
  single: 3.402823E+38
  double: 1.7976931348623157E+308
extended: 1.189731495357231765E+4932
    quad: 1.189731495357231765085759326628007E+4932
epsilon:
  single: 1.192093E-07
  double: 2.2204460492503131E-16
extended: 1.084202172485504434E-19
    quad: 1.925929944387235853055977942584927E-34
troutmask:sgk[278] ldd ./inquiry
./inquiry:
        libgfortran.so.3 => /home/sgk/work/4x/lib/libgfortran.so.3 (0x200748000)
        libm.so.5 => /lib/libm.so.5 (0x200a5d000)
        libgcc_s.so.1 => /home/sgk/work/4x/lib/libgcc_s.so.1 (0x200b80000)
        libc.so.7 => /lib/libc.so.7 (0x200d95000)
        libquad.so.0 => /home/sgk/work/lib/libquad.so.0 (0x200fd6000)



-- 
Steve



More information about the Fortran mailing list