warnings from crayftn

[Knut]

Hi,

currently Ulf and me are sitting in a Cray workshop and tried to compile
FABM with the cray compiler again (and succeeded the first time!). Still
there are 3 warnings left. May the corresponding developers please check
whether the warnings point to critical issues?

Thanks and cheers,
Knut

ftn-7212 crayftn: WARNING SIMPLE_DEPTH_INTEGRAL_DO, File =
../../../../../work/mvkkk258/tools/fabm/code/src/fabm_builtin_models.F90, Line
= 353
Variable "i__" is used before it is defined.

ftn-7212 crayftn: WARNING PHYTO_INTERNAL_NITROGEN, File =
../../../../../work/mvkkk258/tools/fabm/code/src/models/aed/aed_phyto_utils.F90,
Line = 236
Variable "a_nfix" is used before it is defined.


ftn-7206 crayftn: WARNING AED_ZOOPLANKTON_DO_PPDD, File =
../../../../../work/mvkkk258/tools/fabm/code/src/models/aed/aed_zooplankton.F90,
Line = 591
Fortran allocatable variable "prey" is being used before being allocated.

[Karsten Bolding]

Hello

What driver are you using?

If I do:

ifort -I ../include -I ./drivers/gotm/  -P fabm_builtin_models.F90

I get a new file - fabm_builtin_models.i90 with the pre-processed Fortran code for the GOTM physical driver - the offending code - according to crayftn is this:

  subroutine simple_depth_integral_do(self,environment ,fabm_loop_start,fabm_loop_stop )

      class (type_simple_depth_integral),intent(in) :: self

      type (type_environment),intent(inout) :: environment;integer,intent(in) :: fabm_loop_start,fabm_loop_stop ;integer :: i__

      real(rk) :: value,depth

         value = environment%prefetch (i__,self%id_input%index)

as you can see i__ is decleared - can it be the line length in crayftn?

Only looked at this one ....

Karsten

--
You received this message because you are subscribed to the Google Groups "FABM-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fabm-devel+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Karsten Bolding

kar...@bolding-bruggeman.com

+4564422058

[Karsten Bolding]

or do they mean 'initialized' when they write 'defined'?

2015-01-19 15:08 GMT+01:00 Knut <knut...@gmx.net>:

--
You received this message because you are subscribed to the Google Groups "FABM-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fabm-devel+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Knut]

On 01/19/2015 05:33 PM, Karsten Bolding wrote:

or do they mean 'initialized' when they write 'defined'?

The .i90 file and the doc suggest so:

hlogin4:/gfs1/work/mvkkk258/tools/fabm/code/src $ explain ftn-7212

WARNING:  Variable "var" is used before it is defined.

Analysis has found that a variable is used in a statement before any possible
definition of the variable.  This will lead to an unpredictable value being
used in the calculation.

(Unfortunately I did not get "ftn -eP" to work...)

2015-01-19 15:08 GMT+01:00 Knut <knut...@gmx.net>:

Hi,

currently Ulf and me are sitting in a Cray workshop and tried to compile FABM with the cray compiler again (and succeeded the first time!). Still there are 3 warnings left. May the corresponding developers please check whether the warnings point to critical issues?

Thanks and cheers,
Knut

ftn-7212 crayftn: WARNING SIMPLE_DEPTH_INTEGRAL_DO, File = ../../../../../work/mvkkk258/tools/fabm/code/src/fabm_builtin_models.F90, Line = 353
  Variable "i__" is used before it is defined.

ftn-7212 crayftn: WARNING PHYTO_INTERNAL_NITROGEN, File = ../../../../../work/mvkkk258/tools/fabm/code/src/models/aed/aed_phyto_utils.F90, Line = 236
  Variable "a_nfix" is used before it is defined.


ftn-7206 crayftn: WARNING AED_ZOOPLANKTON_DO_PPDD, File = ../../../../../work/mvkkk258/tools/fabm/code/src/models/aed/aed_zooplankton.F90, Line = 591
  Fortran allocatable variable "prey" is being used before being allocated.

--
You received this message because you are subscribed to the Google Groups "FABM-devel" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fabm-devel+...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Karsten Bolding

kar...@bolding-bruggeman.com

+4564422058

--
You received this message because you are subscribed to the Google Groups "FABM-devel" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fabm-devel+...@googlegroups.com.

[Jorn Bruggeman]

I’m aware of the issue affecting fabm_builtin_models. It is real but harmless, as it affects simple_depth_integral_do only in hosts with a depth dimension, and in that case type_simple_depth_integral will never be used (it is simple because it assumes no depth dimension!)

 

Great to hear compilation now succeeds with Cray – what compiler version?

 

Cheers

 

Jorn

[Knut]

Hi Jorn,

I’m aware of the issue affecting fabm_builtin_models. It is real but harmless, as it affects simple_depth_integral_do only in hosts with a depth dimension, and in that case type_simple_depth_integral will never be used (it is simple because it assumes no depth dimension!)

ok.

 

Great to hear compilation now succeeds with Cray – what compiler version?

8.3.5

Cheers,
Knut

[Jorn Bruggeman]

Hi Knut,

 

When I compile GOTM+FABM with Cray 8.3.3 or 8.3.7 I get the error listed below. Since you reported success on Cray, I guess you did not experience this issue. Did you use any special settings? The alternative would be that this issue crept into the code recently, but I’m not aware of significant changes in that routine and also recall seeing this error with earlier Cray compilers… Any ideas?

 

Cheers,

 

Jorn

 

[ 67%] Building Fortran object fabm_src/CMakeFiles/fabm.dir/fabm.F90.o

 

===============================

PDGCS Internal error traceback:

  [0xb19798]]  _expr_type (<unknown>)

  [0x4c932c]]  fei_set_dv_hdr_fld (<unknown>)

  [0x42bf62]]  cvrt_exp_to_pdg (m_cvrt.c:0)

  [0x4333ee]]  cvrt_proc_to_pdg (m_cvrt.c:0)

  [0x4340bc]]  m_cvrt_to_pdg (<unknown>)

  [0x4574ca]]  m_start_ipa (<unknown>)

  [0x415d6d]]  main (<unknown>)

 

===============================

Note:  This is a non-debug compiler.  Technical support should

       continue problem isolation using a compiler built for

       debugging.

 

===============================

Src_path_name: /home1/n01/n01/jbr/fabm-ersem/src/fabm.F90

Src_file_name: /home1/n01/n01/jbr/fabm-ersem/src/fabm.F90

Routine_name : classify_variables$fabm_

Target.name  : x86-64

ftn-7991 crayftn: INTERNAL CLASSIFY_VARIABLES, File = ../../../home1/n01/n01/jbr/fabm-ersem/src/fabm.F90, Line = 3507

  INTERNAL COMPILER ERROR:  "_expr_type: Invalid table type" (v_expr_utl.c v95518, line 6786)

make[3]: *** [fabm_src/CMakeFiles/fabm.dir/fabm.F90.o] Error 1

make[2]: *** [fabm_src/CMakeFiles/fabm.dir/fabm.F90.o.provides] Error 2

make[1]: *** [fabm_src/CMakeFiles/fabm.dir/all] Error 2

make: *** [all] Error 2

[Knut]

Dear Jorn,

I only "succeeded" with -O0...
Higher optimisation levels also caused an internal compiler error. I forwarded to the Cray guys and they fixed it in the upcoming compiler version.

Cheers,
Knut

[Jorn Bruggeman]

Hi Knut,

 

Confirmed – compiles here with -O0 too. Thanks!