!     pgfortran -acc -Mcuda -ta=nvidia,cc20 pi-pgicu-reduce-f.f -o pi-pgicu-reduce-f
      program main
      integer n, i
      double precision  PI25DT
      parameter (PI25DT = 3.141592653589793238462643d0)
      double precision  pi, h, sum, x, f, a
      real difftime, t(2)

!                                 function to integrate
      f(a) = 4.d0 / (1.d0 + a*a)
 10   write(6,98)
 98   format('Enter the number of intervals: (0 quits)')
      read(5,99) n
 99   format(i10)
      difftime = dtime(t)
!                             check for quit signal
      if ( n .le. 0 ) goto 30
!                    calculate the interval size
      h = 1.0d0/n
      sum  = 0.0d0
!$acc parallel reduction(+:sum) private(i,x)
      do i = 1, n, 1
        x   = h * (dble(i) - 0.5d0)
        sum = sum + f(x)
      enddo
!$acc end parallel
      pi = h * sum
      difftime = dtime(t)
      write(6, 96) pi, abs(pi - PI25DT)
 96   format('pi is approximately: ', F18.16, ', Error is: ', F18.16)
      write(6, 97) difftime
 97   format('wall clock time = ', F18.6)
      goto 10
 30   end