DT and NDTFAST trouble

[langlangxie]

hello,everyone!
I have a trouble in my roms.in.
the first version:
NTIMES == 1036800 ！50 years
DT == 600.0d0
NDTFAST == 180

NRREC == 0
LcycleRST == T
NRST == 51840
NSTA == 1
NFLT == 1
NINFO == 51840

LDEFOUT == T
NHIS == 0
NDEFHIS == 0
NTSAVG == 1
NAVG == 4320
NDEFAVG == 51840
NTSDIA == 1
NDIA == 0
NDEFDIA == 0

LcycleTLM == F
NTLM == 720
NDEFTLM == 0
LcycleADJ == F
NADJ == 720
NDEFADJ == 0
NSFF == 720
NOBC == 720
In the version,the ROMS run good.
Then I have change my DT and NDTFAST,as
NTIMES == 1555200
DT == 200.0d0
NDTFAST == 30

NRREC == 0
LcycleRST == T
NRST == 155520
NSTA == 1
NFLT == 1
NINFO == 1

LDEFOUT == T
NHIS == 0
NDEFHIS == 0
NTSAVG == 1
NAVG == 129600
NDEFAVG == 155520
NTSDIA == 1
NDIA == 0
NDEFDIA == 0

LcycleTLM == F
NTLM == 2160
NDEFTLM == 0
LcycleADJ == F
NADJ == 2160
NDEFADJ == 0
NSFF == 2160
NOBC == 2160
The roms have a error :
[node53:mpi_rank_36][error_sighandler] Caught error: Segmentation fault (signal 11)

so can you help me to solve the trouble?
NTIMES=YEAR(20 YEARS)*MONTH(30 DAYS)*DAY(24 HOURS)*HOUR(60 MINUTES)*MINUTE（60 SECONDS）/DT，is right?
Thanks in advance!

[rsoutelino]

Did you try to increase the stack limit in the machine you're running? Just a guess, but since you decreased the time step, there is more memory being used in between netcdf time records.

Code: Select all

ulimit -s unlimited

[langlangxie]

Did you try to increase the stack limit in the machine you're running? Just a guess, but since you decreased the time step, there is more memory being used in between netcdf time records.

Code: Select all

ulimit -s unlimited

Thanks,can you tell me how to find ulimit.I change my NRST,NAVG and so on.
thanks again!

[kate]

Ulimit is a built-in operation in the bash family of Unix shells. It isn't in the ROMS input file, but might work from the job script (if you use a job script). One of the many ways to get a seg fault is to use too much memory - but it isn't the only way. I can't guess how you got yours without say recompiling for debugging then running in a debugger. Did it happen right away or after many time steps of running?