* gfortran6 with OpenACC - error
@ 2016-04-26 15:08 Anton Shterenlikht
2016-04-26 15:16 ` Cesar Philippidis
0 siblings, 1 reply; 3+ messages in thread
From: Anton Shterenlikht @ 2016-04-26 15:08 UTC (permalink / raw)
To: fortran
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1305 bytes --]
gfortran 6.0.0 20160410 on linux.
Compiling with *no* -fopenacc works fine,
and the serial program runs as expected.
Compiling with -fopenacc gives an error.
Am I using OpenACC directives wrongly?
newblue4> cat z.f90
program z
implicit none
integer, parameter :: d1 = 1000, d2 = 1000
real :: tmp( (d1-2) *(d2-2) )
real, allocatable :: arr(:,:), old(:,:)
integer :: i,j,k
! allocate and set all to zero
allocate( arr(d1,d2), source = 0.0 )
allocate( old(d1,d2) )
! set inner cells to random numbers
call random_number(tmp)
arr( 2:d1-1 , 2:d2-1 ) = reshape( tmp, (/ d1-2, d1-2 /) )
!$acc data copy(arr)
write (*,*) "before:", sum(abs(arr))
do i=1,1000
old = arr
!$acc kernels
do j=2,d1-1
do k=2,d2-1
arr(j,k) = 0.25 * ( old(j-1,k) + old(j+1,k) + old(j,k-1) + old(j,k+1) )
end do
end do
!$acc end kernels
end do
write (*,*) "after:", sum(abs(arr))
end program z
newblue4>
newblue4> gfortran -O2 z.f90 -Wall -o gfort-acc.x
newblue4> time ./gfort-acc.x
before: 497871.688
after: 464004.844
9.102u 0.006s 0:09.13 99.6% 0+0k 0+0io 0pf+0w
newblue4>
newblue4> gfortran -fopenacc -O2 z.f90 -Wall -o gfort-acc.x
z.f90:34:3:
end program z
1
Error: Unexpected END statement at (1)
f951: Error: Unexpected end of file in âz.f90â
newblue4>
Thanks
Anton
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: gfortran6 with OpenACC - error
2016-04-26 15:08 gfortran6 with OpenACC - error Anton Shterenlikht
@ 2016-04-26 15:16 ` Cesar Philippidis
2016-04-26 15:37 ` Anton Shterenlikht
0 siblings, 1 reply; 3+ messages in thread
From: Cesar Philippidis @ 2016-04-26 15:16 UTC (permalink / raw)
To: mexas, fortran
On 04/26/2016 08:08 AM, Anton Shterenlikht wrote:
> gfortran 6.0.0 20160410 on linux.
> Compiling with *no* -fopenacc works fine,
> and the serial program runs as expected.
> Compiling with -fopenacc gives an error.
> Am I using OpenACC directives wrongly?
>
> newblue4> cat z.f90
> program z
> implicit none
> integer, parameter :: d1 = 1000, d2 = 1000
> real :: tmp( (d1-2) *(d2-2) )
> real, allocatable :: arr(:,:), old(:,:)
> integer :: i,j,k
>
> ! allocate and set all to zero
> allocate( arr(d1,d2), source = 0.0 )
> allocate( old(d1,d2) )
>
> ! set inner cells to random numbers
> call random_number(tmp)
> arr( 2:d1-1 , 2:d2-1 ) = reshape( tmp, (/ d1-2, d1-2 /) )
>
> !$acc data copy(arr)
> write (*,*) "before:", sum(abs(arr))
>
> do i=1,1000
> old = arr
>
> !$acc kernels
> do j=2,d1-1
> do k=2,d2-1
> arr(j,k) = 0.25 * ( old(j-1,k) + old(j+1,k) + old(j,k-1) + old(j,k+1) )
> end do
> end do
> !$acc end kernels
>
> end do
>
> write (*,*) "after:", sum(abs(arr))
>
> end program z
> newblue4>
>
> newblue4> gfortran -O2 z.f90 -Wall -o gfort-acc.x
> newblue4> time ./gfort-acc.x
> before: 497871.688
> after: 464004.844
> 9.102u 0.006s 0:09.13 99.6% 0+0k 0+0io 0pf+0w
> newblue4>
>
> newblue4> gfortran -fopenacc -O2 z.f90 -Wall -o gfort-acc.x
> z.f90:34:3:
>
> end program z
> 1
> Error: Unexpected END statement at (1)
> f951: Error: Unexpected end of file in ‘z.f90’
> newblue4>
This program missing an "!$acc end data" to mark the end of the data region.
Cesar
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: gfortran6 with OpenACC - error
2016-04-26 15:16 ` Cesar Philippidis
@ 2016-04-26 15:37 ` Anton Shterenlikht
0 siblings, 0 replies; 3+ messages in thread
From: Anton Shterenlikht @ 2016-04-26 15:37 UTC (permalink / raw)
To: cesar, fortran, mexas
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 426 bytes --]
>> newblue4> gfortran -fopenacc -O2 z.f90 -Wall -o gfort-acc.x
>> z.f90:34:3:
>>
>> end program z
>> 1
>> Error: Unexpected END statement at (1)
>> f951: Error: Unexpected end of file in âz.f90â
>> newblue4>
>
>This program missing an "!$acc end data" to mark the end of the data region.
>
>Cesar
Thank you
The error text made me think something
was wrong with my Fortran statements,
not with OpenACC.
Anton
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2016-04-26 15:37 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-04-26 15:08 gfortran6 with OpenACC - error Anton Shterenlikht
2016-04-26 15:16 ` Cesar Philippidis
2016-04-26 15:37 ` Anton Shterenlikht
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).