some error when run example/GLBT0.08/expt_73.7

[xing cong]

Dear HYCOM Support Team,

I hope this message finds you well.

I have get this data successfully, I have compiled the src code with intel MPI and openMP and I have writed a slurm script to submit a job, but I meet a error when I run this progress with these data, can you help me?

the error is :

**************************************************

xcspmd: patch.input for wrong nreg

**************************************************

this is error img:

the patch.input file:

[Alan Wallcraft]

Support for global tripole (Arctic bi-pole) grids has to be enabled at compile time.  Set OCN_GLB in   Make.csh

# Global or regional
#setenv OCN_GLB -DARCTIC ## global tripolar simulation
setenv OCN_GLB ""

Then delete everything from the old make:

/bin/rm *.o *.mod hycom *.log

and remake.

If you are going to run both global and regional models then you need two executables and hence two src directories.

Alan.

[xing cong]

Thank you  very much, the error is sloved~

But now I have another error, because my major is computer science, so I can not understand some terms in HYCOM. 🙏🙏

I am very need to run HYCOM in my HPC cluster, 😭

the error is ：

tsofrq = 180

error in blkini - input tsofrq but should be mtracr

related code in the file blkdat.F90 : 

      if (mnproc.eq.1) then
      write(lp,*)
      endif !1st tile
      call blkini(tsofrq,'tsofrq')
      call blkinr(tofset,'tofset','(a6," =",f10.4," degC/century")')
      call blkinr(sofset,'sofset','(a6," =",f10.4,"  psu/century")')

[Alan Wallcraft]

There are two new blkdat.input entries in the latest version of HYCOM (mtracr and lbmont) that are missing from HYCOM-examples:

   0      'trcflg' = tracer flags      (one digit per tr, most sig. replicated)
   0      'mtracr' = number of diagnostic tracers

   0      'lbflag' = lateral barotropic bndy flag (0=none, 1=port, 2=input)
   0      'lbmont' = baro nesting archives have sshflg=2 (0=F,1=T)

Add these lines and the case should run.

Alan.

[xing cong]

Thanks, it is very help for the error, just add  ` 0      'mtracr' = number of diagnostic tracers`  line in blkdat.input , the error can be sloved. 

Now, I can run this example for about 40 min, but it get another error 😂, the error is :

31174470 (2019/198 21) region-wide mean Density Dev: 2.6279637576

error in zaiowr - can't write record 96 on array I/O unit 13.

mpierr = 268493344

**************************************************

(zaiowr)

**************************************************

XCHALT CALLED ON PROC = 894 1 22

**************************************************

Abort(9) on node 893 (rank 893 in comm 496): application called MPI_Abort(comm=0x84000002, 9) - process 893

mlx5: h07r2n13: got completion with error:

00000000 00000000 00000000 00000000

00000000 00000000 00000000 00000000

00000011 00000000 00000000 00000000

00000000 00008914 0a014211 38a99dd2

[h07r2n13:21971:0:21971] Caught signal 11 (Segmentation fault: address not mapped to object at address 0x168)

I am very grateful for your previous assistance and hope you can help me again ❤️❤️.

[xing cong]

Dear HYCOM Support Team, I hope this message finds you well. 

The problem is solved, it occurs due to insufficient memory to store data. (can't write record 96 on array I/O unit 13.)

Currently, I can run the example for about an hour.

When it finishes, I receive this log. Does this mean I have successfully run the code and simulated the data?

step 31174560 day 43298.00000 -- archiving completed --

timer statistics, processor 293 out of 1525

-----------------------------------------------

xcaget calls = 2324 time = 5.87015 time/call = 0.00252588

xcaput calls = 330 time = 0.22571 time/call = 0.00068396

xcsum calls = 3758 time = 197.19908 time/call = 0.05247448

xcmaxr calls = 7047 time = 320.04184 time/call = 0.04541533

xctilr calls = 83706 time = 1149.04671 time/call = 0.01372717

zaio** calls = 68 time = 7.11530 time/call = 0.10463669

zaiord calls = 330 time = 76.45293 time/call = 0.23167556

zaiowr calls = 2324 time = 444.27128 time/call = 0.19116664

xc**** calls = 1 time = 3342.55012 time/call = 3342.55011995

cnuity calls = 720 time = 256.02532 time/call = 0.35559072

tsadvc calls = 720 time = 268.26082 time/call = 0.37258448

momtum calls = 720 time = 402.06764 time/call = 0.55842727

barotp calls = 720 time = 593.22508 time/call = 0.82392372

thermf calls = 720 time = 28.88789 time/call = 0.04012208

ic**** calls = 720 time = 140.44647 time/call = 0.19506454

mx**** calls = 720 time = 383.37529 time/call = 0.53246569

conv** calls = 720 time = 0.00005 time/call = 0.00000006

diapf* calls = 720 time = 0.00005 time/call = 0.00000006

hybgen calls = 720 time = 167.46184 time/call = 0.23258588

overtn calls = 1 time = 1.65750 time/call = 1.65749601

archiv calls = 25 time = 451.80732 time/call = 18.07229282

incupd calls = 720 time = 0.00037 time/call = 0.00000052

aslsav calls = 720 time = 20.92804 time/call = 0.02906672

asseln calls = 720 time = 26.02908 time/call = 0.03615149

total calls = 1 time = 3031.60495 time/call = 3031.60495153

processor 1: memory (words) now,high = 28644078 28644078

processor 1: memory (GB) now,high = 0.213 0.213

processor 1: eq. 3-D arrays now,high = 75.479 75.479

processor 12: memory (words) now,high = 28644078 28644078

processor 12: memory (GB) now,high = 0.213 0.213

processor 12: eq. 3-D arrays now,high = 75.479 75.479

processor 1525: memory (words) now,high = 28147522 28147522

processor 1525: memory (GB) now,high = 0.210 0.210

processor 1525: eq. 3-D arrays now,high = 74.171 74.171

**************************************************

(normal)

**************************************************

[Alan Wallcraft]

There is an accuracy criteria in README.GLBT0.08_737 but since you are not using this for a formal benchmark you have successfully run the code.

However, 50 minutes is 5x slower than the EXAMPLE (which is on a HP Cray EX system, i.e. a high end supercomputer).

narwhal08 203> ../total.csh
HY01525.log:   total    calls =        1   time =  623.0 10.38 wall mins

The I/O is 19x slower:

narwhal08 209> ../zaio.csh
HY01525.log:   zaiowr   calls =     2324   time =   23.3

This might be right.  One way to check is to try running on other core counts and confirm that the scaling is good (constant corehrs is perfect scaling).

Alan.

[xing cong]

Dear Professor,

I hope this message finds you well. I would like to extend my sincere gratitude for your invaluable assistance in the past. Your guidance has been instrumental in my research endeavors.

I am currently facing an issue with running an OMPI-type program on a platform that exclusively uses the GNU compiler. Previously, I successfully executed similar programs on a platform equipped with the Intel compiler and OMPI. However, upon transitioning to the GNU compiler environment, I encountered the following error:

Error termination. Backtrace:

At line 2453 of file blkdat.F90 (unit = 2099, file = './blkdat.input')

Fortran runtime error: Bad integer for item 1 in list input

FC = mpifort

FCFFLAGS = -fbacktrace -march=native -O2 -fdefault-real-8 -ffloat-store -fopenmp -w -mcmodel=small

-fdefault-double-8 -fPIC -fno-second-underscore

CC = mpicc

CCFLAGS = -O

CPP = cpp -P

CPPFLAGS = -DREAL8 -DMPI -DSERIAL_IO -DENDIAN_IO -DNAN2003 -DTIMER -DRELO $(CPP_EXTRAS)

LD = $(FC)

LDFLAGS = $(FCFFLAGS)

EXTRALIBS =

Thank you once again for your time and consideration. I look forward to your response.

Warm regards,

CongXing

[Alan Wallcraft]

Please provide the last few lines of model output before the model dies.  These will have been printed out after reading them from blkdat.input, e.g.:

iversn =        22
iexpt  =       998
 
idm    =       500
jdm    =       382
 

.....

incflg =         0
incstp =        12
incupf =         1
 
ra2fac =    0.1250
wbaro  =    0.1250
btrlfr =         T
btrmas =         F

...

relax  =         F
trcrlx =         F
priver =         F
epmass =         F

Also send the blkdat.input lines around the last few in the printout. The error is probably for the blkdat variable immediately after the last one in the printout.

Alan.