1! { dg-do link } 2! 3! This test checks whether the largest possible 4! floating-point number works. That's usually 5! REAL(16) -- either because the hardware supports it or 6! because of libquadmath. However, it can also be 7! REAL(10) or REAL(8) 8! 9program test_qp 10 use iso_fortran_env, only: real_kinds 11 implicit none 12 integer, parameter :: QP = real_kinds(ubound(real_kinds,dim=1)) 13 real(QP), parameter :: Z1 = 1,HALF_PI = asin(Z1),PI = HALF_PI+HALF_PI 14 real(QP) :: x = 0.124_QP 15 complex(QP) :: z = 0.124_QP 16 print *, 'kind = ', qp 17 print *, x 18 print *, PI 19 print *, 16*atan(0.2_QP)-4*atan(Z1/239) 20 print *, sin(PI) 21 print *, cos(HALF_PI) 22 print *, asinh(PI) 23 print *, erfc(Z1) 24 print *, epsilon(x) 25 print *, precision(x) 26 print *, digits(x) 27 28 print *, z 29 print *, PI*cmplx(0.0_qp, 1.0_qp) 30! Disable the complex functions as not all "long-double" systems have 31! a libm with those C99 functions. (libquadmath had), cf. PR 46584 32! print *, 16*atan(0.2_QP)-4*atan(Z1/239) 33! print *, sin(z) 34! print *, cos(z) 35! print *, sinh(z) ! asinh not implemented in libquadmath, cf. PR 46416 36 print *, precision(z) 37end program test_qp 38