------- Comment #8 from gdsjaar at sandia dot gov 2009-07-07 20:13 -------
Subject: Re: sign intrinsic fails for value of 0.0
OK, sorry for the buggy example code. It illustrates the folly of
improving the code after doing the runs. A better fixed code
illustrating my point is:
program main
val = 0.0
test = sign(0.5, val) - sign(0.5, -val)
test2 = ysign(0.5, val) - ysign(0.5, -val)
if (test .ne. test2) then
write (*,*) 'fail'
else
write (*,*) 'pass'
end if
stop
end
real function ysign(a, b)
C ...Returns `ABS(A)*s', where s is +1 if `B.GE.0', -1 otherwise.
if (b .ge. 0) then
s = 1.0
else
s = -1.0
end if
ysign = abs(a) * s
return
end
kargl at gcc dot gnu dot org wrote:
> ------- Comment #6 from kargl at gcc dot gnu dot org 2009-07-07 20:05 -------
> (In reply to comment #3)
>
>
>> OK, so I should instead be submitting a bug report for intel and g77 and
>> pgi. gfortran is the only correct implementation?
>>
>
> g77 is no longer supported. You can do want you want with bug
> reports to intel and pgi. But you should fix your code. Here's
> the output with a few PRINTs added to your original code.
>
> REMOVE:kargl[214] ./z
> 0.50000000 -0.50000000
> 1.0000000 1.05696461E+09
> fail
>
> Here's the output after I fixed your code.
> REMOVE:kargl[216] ./z
> 0.50000000 -0.50000000
> 1.0000000 0.50000000
> fail
>
> Here's the fixed code.
>
> program main
> real, external :: mysign
> val = 0.0
> test = sign(0.5, val) - sign(0.5, -val)
> test2 = mysign(0.5, val)
>
> print *, sign(0.5, val), sign(0.5, -val)
> print *, test, test2
>
> if (test .ne. test2) then
> write (*,*) 'fail'
> else
> write (*,*) 'pass'
> end if
> stop
> end
>
> real function mysign(a, b)
> C ...Returns `ABS(A)*s', where s is +1 if `B.GE.0', -1 otherwise.
> if (b .ge. 0) then
> s = 1.0
> else
> s = -1.0
> end if
> mysign = abs(a) * s
> return
> end
>
>
>
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40675
