PGI User Forum
 SearchSearch   MemberlistMemberlist     RegisterRegister   ProfileProfile    Log inLog in 

CUDA-x86.

error with directives and clauses

 
Post new topic   Reply to topic    PGI User Forum Forum Index -> Accelerator Programming
View previous topic :: View next topic  
Author Message
Tuan



Joined: 11 Jun 2009
Posts: 226
PostPosted: Sat Jan 23, 2010 7:55 pm    Post subject: error with directives and clauses Reply with quote
My compile:
pgf95 main.f -ta=nvidia -Minfo -Mextend -Mpreprocess

My compiling error:
PGF90-S-0000-Internal compiler error. semsmp: bad rednum 1204 (main.f: 105)
0 inform, 0 warnings, 1 severes, 0 fatal for foofunc

Code:


#ifdef _ACCEL
      use accel_lib
#endif

     
      call  foofunc()

      end

!===================================================================
!===================================================================
      subroutine foofunc()

      IMPLICIT double precision (a-h,o-z)
      PARAMETER (maxiter=10000000,noutincr=100)
      PARAMETER (nouter=maxiter/noutincr)
      DIMENSION ca_ds(N), ca_jsr(N),c(N)
      DIMENSION aJ_ryr(N)
      DIMENSION aJ_efflux(N),aJ_refill(N)
      DIMENSION bigC(N)
      DIMENSION chiC(N),chiO(N)
      DIMENSION compKm(irow_R,0:maxnklm), indexKm(irow_R,maxnklm)
      DIMENSION compKp(irow_R,0:maxnklp), indexKp(irow_R,maxnklp)
      double precision, dimension(irow_R)::  U_Ro
      double precision, dimension(N) :: ryrgate
      DIMENSION isfu(N)
!local
      INTEGER LSTATE,N
      INTEGER I,INFO,NSKIP,SEED(LSEED),STATE(LSTATE)
      double precision A,B, X(N)
      double precision, dimension(N, maxnklm+maxnklp-1) :: compPdt
      integer, dimension(N, maxnklm+maxnklp-1) :: indexPdt
      double precision, dimension(0:maxnklm+maxnklp-1,N)  ::  compP
      double precision :: Pdtmin
      integer, dimension(N) :: flag

      irow_R = 1000
      Mbig=irow_R

      isfu = 1

     
      ryrgate(:) = U_Ro(isfu(:))

      dt = 1d-7 ! seconds
      Vstep = -10d0
      time = 0d0         ! seconds
      timeend = 1d0      ! seconds
     
      compP(0,:) = +0.0
!$acc data region copyin(aKo, aK1, ryrgate, compKp, indexKp, compKm, indexKm, isfu, ca_ds, ca_jsr, compP, U_Ro),
!$acc$local(bigC, chiC, chiO, compPdt, indexPdt)     
      do iouter=1,nouter
       do iinner=1,noutincr
         Pdtmin = 0.0d0
!$acc region
         do i=1, maxnklm+maxnklp-1
            do j = 1,N
               compPdt(j,i) = 0.0d0
               indexPdt(j,i) = 0
            enddo
         enddo
!$acc end region         
         
!$acc region
    do i=1,N
       bigc(i) = ca_ds(i)**4/ ((aKo - aK1*ca_jsr(i))**4. + ca_ds(i)**4)
       chiC(i) = 1d0 + (ryrgate(i))/N_R
       chiO(i) = 1d0 + (N_R-ryrgate(i))/N_R
         enddo
!$acc end region


         dtmax =-0.1/Pdtmin
         dt = min(dtmax,1d-5)
         time = time + dt

!$acc region local(flag, aJ_ryr, aJ_efflux, aJ_refill)
         do i=1,N
            compPdt(i, 1) = compPdt(i, 1) + 1/dt
            flag(i) = 1
          enddo
!$acc end region

         
       enddo !end inner time loop
!$acc update host(ca_ds)
      enddo !end outer time loop

   90 continue

!$acc end data region
      return
      end


Q1: could someone tell me why I get the error above?
Q2: using !$acc data region directive to wrap this code
Code:
         
         dtmax =-0.1/Pdtmin
         dt = min(dtmax,1d-5)
         time = time + dt

would it be executed on Accelerator while it is not enclosed by any compute region?

Thanks,
Tuan
Back to top
View user's profile
mkcolg



Joined: 30 Jun 2004
Posts: 4996
Location: The Portland Group Inc.
PostPosted: Mon Jan 25, 2010 9:59 pm    Post subject: Reply with quote
Hi Tuan,

Quote:
Q1: could someone tell me why I get the error above?


This appears to be a compiler error. I have added a technical problem report (TPR#16535) and sent it to our compiler engineers. Note that this is a severe error so I have given it a very high priority.

Quote:
Q2: using !$acc data region directive to wrap this code
Code:

dtmax =-0.1/Pdtmin
dt = min(dtmax,1d-5)
time = time + dt


would it be executed on Accelerator while it is not enclosed by any compute region?


No, this section of code would execute on the host. Data region only control when date is copied to and from the GPU..

Thanks,
Mat
Back to top
View user's profile
jtull



Joined: 30 Jun 2004
Posts: 234
PostPosted: Wed Mar 17, 2010 3:14 pm    Post subject: TPR 16528 Scoping issue causes syntax error Reply with quote
We have corrected this problem in the current 10.3 release.

Thanks again for your submission
Back to top
View user's profile
jtull



Joined: 30 Jun 2004
Posts: 234
PostPosted: Thu Mar 18, 2010 8:01 am    Post subject: TPR 16535 ACC: User code causes pgf901 seg fault Reply with quote
The problem reported in this forum as TPR 16535 has been corrected in the 10.3 release.
Thanks again for your submission.

regards,
dave
Back to top
View user's profile
Display posts from previous:   
Post new topic   Reply to topic    PGI User Forum Forum Index -> Accelerator Programming All times are GMT - 7 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2002 phpBB Group