IFX vs IFORT - crash

[joe-griffin]

I am getting a crash when one routine accesses a module with IFX on linux and windows.  This code works fine with IFORT.

The MODULE source, DBCXXX_mod.F90,   has:

module DBCXXX_mod

…

      INTEGER  ZQZQZQ, DBCXXX_IZ

…

      POINTER ( DYNWORKOFF4PTR, DBCXXX_IZ(*) )

The calling routine, bulfun.F90, has:

      use DBCXXX_mod

…

      Logical chk1, chk2, chk3

      integer ifpstar,iextra

…

            chk1 = DBCXXX_iz(n+ii).eq.DBCXXX_iz(n+ii+1)

            chk2 = iextra.eq.DBCXXX_iz(n+ii)

            chk3 = iextra.eq.DBCXXX_iz(n+ii+1)

!           if ( chk1 .or. chk2 .or. chk3) then

            if ( chk3 ) then   ! chk1 is ok, but chk2 or chk3 cause CRASH .. This code is NEVER executed btw.

On WINDOWS I see the problem with the 2024.2 compilers.

On LINUX I still am using the 2023.1 compilers, but suspect that the problem is with 2024.2 since I see the same behavior on my window’s system.   I hope to get access to the 2024.2 linux compilers next week.

I compile on linux with:

/opt/intel/oneapi/compiler/2023.1.0/linux/bin/ifx -c -o /scratch/jjg/xxx_icx/LX8664_SOURCE_OPT/utilities/non_deliverable/sdl/sdl_modules/DBCXXX_mod.o -nbs -pad-source -save -zero -warn nousage -WB -W0 -Qdyncom XNSTRN -mp1 -pc80 -O2 -integer-size 64 -real-size 64 -double-size 128 -mcmodel=large -O2 -g -traceback -shared-intel -debug inline-debug-info -parallel-source-info=2 -fPIC -I/opt/intel/oneapi/compiler/2023.1.0/linux/compiler/include -I/opt/intel/oneapi/compiler/2023.1.0/linux/compiler/include/intel64 -I/scratch/jjg/xxx_icx/LX8664/include -I/scratch/jjg/xxx_icx/IMPORTS/scasystem/LX8664/include -module /scratch/jjg/xxx_icx/LX8664_SOURCE_OPT/utilities/non_deliverable/sdl/modules /scratch/jjg/xxx_icx/LX8664_SOURCE_OPT/utilities/non_deliverable/sdl/sdl_modules/DBCXXX_mod.f90

And:

/opt/intel/oneapi/compiler/2023.1.0/linux/bin/ifx -c -o /scratch/jjg/xxx_icx/LX8664_SOURCE_OPT/utilities/non_deliverable/sdl/indep/bulfun.o -nbs -pad-source -save -zero -warn nousage -WB -W0 -Qdyncom XNSTRN -mp1 -pc80 -O2 -integer-size 64 -real-size 64 -double-size 128 -mcmodel=large -O2 -g -traceback -shared-intel -debug inline-debug-info -parallel-source-info=2 -O0 -unroll0 -fno-vec-peel-loops -fno-vec-remainder-loops -fPIC -I/opt/intel/oneapi/compiler/2023.1.0/linux/compiler/include -I/opt/intel/oneapi/compiler/2023.1.0/linux/compiler/include/intel64 -I/scratch/jjg/xxx_icx/LX8664/include -I/scratch/jjg/xxx_icx/IMPORTS/scasystem/LX8664/include -module /scratch/jjg/xxx_icx/LX8664_SOURCE_OPT/utilities/non_deliverable/sdl/modules /scratch/jjg/xxx_icx/LX8664_SOURCE_OPT/utilities/non_deliverable/sdl/indep/bulfun.f90

Should I change compile options in either the MODULE compile or the calling routine?

[andrew_4619]

It is hard to draw any conclusion from the code snippets. Often such snippets miss showing the vital ingredient of the failure recipe. The best way is to create a small program that demonstrated the problem and then it can be more readily diagnosed. 

[Ron_Green]

Don't bother with 2024.2.  version 2025.0.0 will or is available now.  2023.1 is not nearly as stable as 2024.2.1 or 2025.0.0