Segmentation fault on both CPU and GPU builds

[TaufeqRazakh]

I tried a CPU build with the following recipe

cmake -DCMAKE_CXX_COMPILER=icpx -DCMAKE_Fortran_COMPILER=mpif90 ..

DAT(input) file was built once with the following command

cd init
make all
/geninit -i nh3.xyz -n 
./geninit -i norm.xyz 
cp rxff.bin ../DAT

And I run into the following error when using the executble from the project dir where DAT/ is

~/RXMD> ./rxmd_exe 
INFO: myid,hostname          0 x1921c0s
              rxmd has started
================================================================================
element,mass,filename:   1-H      1.000   pot/deployed.pth
element,mass,filename:   2-N     14.000   pot/deployed.pth
================================================================================
 get_mdcontext_func : mdcontext_rxmdnn
================================================================================
element,mass,filename:   1-H      1.000   pot/deployed.pth
element,mass,filename:   2-N     14.000   pot/deployed.pth
================================================================================
INFO: Opening file in ReadBIN(): DAT/rxff.bin
------------------------------------------------------------
         req/alloc # of procs:        1  /        1
         req proc arrangement:        1        1        1
                time step[fs]:    2.00E-01
 MDMODE CURRENTSTEP NTIMESTEP:  1         0     20000
treq,vsfact,sstep,vmag_factor:      10.000   0.990       10   5.000
    fstep,pstep,xyz_num_stack:   100    10     1
               NATOMS GNATOMS:                     432                     432
                         LBOX:       1.000       1.000       1.000
                  Hmatrix [A]:         16.515          0.000          0.000
                  Hmatrix [A]:          0.000         16.515          0.000
                  Hmatrix [A]:          0.000          0.000         16.515
               lata,latb,latc:      16.515      16.515      16.515
          lalpha,lbeta,lgamma:      90.000      90.000      90.000
          NBUFFER, MAXNEIGHBS:      100000         100
------------------------------------------------------------
                     DataDir :         DAT
             FFPath, ParmPath:      ffield      rxmd.in
------------------------------------------------------------
------------------------------------------------------------
               density [g/cc]:    0.6768
         # of linkedlist cell:******************
            maxrc, lcsize [A]:     0.000       -0.00     -0.00     -0.00
          # of atoms per type:         324 - 1         108 - 2 
------------------------------------------------------------
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source             
libze_intel_gpu.s  000014908752C4D5  Unknown               Unknown  Unknown
libpthread-2.31.s  00001490909BC8C0  Unknown               Unknown  Unknown
rxmd_exe           0000000000417273  Unknown               Unknown  Unknown
rxmd_exe           00000000004691E5  Unknown               Unknown  Unknown
rxmd_exe           00000000004695E6  Unknown               Unknown  Unknown
rxmd_exe           000000000040CE5C  Unknown               Unknown  Unknown
rxmd_exe           000000000040CC6D  Unknown               Unknown  Unknown
libc-2.31.so       00001490907E624D  __libc_start_main     Unknown  Unknown
rxmd_exe           000000000040CB9A  Unknown               Unknown  Unknown

Any help would be greatly appreciated.

[ye-luo]

compiler with -g and run with a debugger.

[TaufeqRazakh]

Looks like the issue is from a bad division when adding a breakpoint to

• utils::update_box_params

(gdb) list - 
271	LBOX(1)=lata/vprocs(1)
272	LBOX(2)=latb/vprocs(2)
273	LBOX(3)=latc/vprocs(3)
274	
275	!--- get the number of linkedlist cell per domain
276	cc(1:3)=int(LBOX(1:3)/maxrc)
277	
278	!--- local system size in the unscaled coordinate.
279	LBOX(1:3) = 1.d0/vprocs(1:3)
280	
(gdb) p vprocs 
$1 = (1, 1, 1)
(gdb) p lata
$2 = 16.51501
(gdb) p lbox 
$3 = (7.1253040736177892e-317, 7.1258139493642974e-317, 2.3815782291124779e-317)