Re: Segmentation fault (core dumped)

Hi Vishali

If your files have one year of 4x daily data,
you have 4*365*17*73*144 = 260,907,840 numbers on each variable
stored there.
If they're floating point (4 bytes) this is ~ 1GB,
if they're double (8 bytes) this is ~ 2GB.
You may be on the bleeding edge with variable
w = f->WSpeed.
You also need memory for the other variables (wMon, etc), for the OS,
the NCL executable, and so on.

I wonder if the problem is this.
How much memory does your computer have?
Is it a 32- or 64-bit machine?
NCL 5.2 only supports variables up to 2GB large.
NCL 6.0 supports up to the physical RAM available,
*if* your computer is 64-bit.

My two cents,
Gus Correa

Vishali P. wrote:
> Hi
>
>
> [ 1 ] Output of first separated script
>
> Variable: nfilu
> Type: integer
> Total Size: 4 bytes
> 1 values
> Number of Dimensions: 1
> Dimensions and sizes: [1]
> Coordinates:
> (0) 7
>
>
> Variable: filu
> Type: string
> Total Size: 28 bytes
> 7 values
> Number of Dimensions: 1
> Dimensions and sizes: [7]
> Coordinates:
> (0) wind.1998.nc <http://wind.1998.nc>
> (1) wind.1999.nc <http://wind.1999.nc>
> (2) wind.2000.nc <http://wind.2000.nc>
> (3) wind.2001.nc <http://wind.2001.nc>
> (4) wind.2002.nc <http://wind.2002.nc>
> (5) wind.2003.nc <http://wind.2003.nc>
> (6) wind.2004.nc <http://wind.2004.nc>
>
>
>
> [ 2 ] Output of second separated Script
>
> (0) success
>
>
>
> [ 3 ] Output ncl_filedump of 1st file ( wind speed of 1998), i got :
>
> Variable: f (file variable)
>
> filename: wind.1998
> path: wind.1998.nc <http://wind.1998.nc>
> file global attributes:
> CDI : Climate Data Interface version 1.4.1
> Conventions : CF-1.0
> history : Mon Oct 18 11:24:43 2010: cdo mergetime wind1998a.nc
> <http://wind1998a.nc> wind1998b.nc <http://wind1998b.nc> wind.1998.nc
> <http://wind.1998.nc>
> CDO : Climate Data Operators version 1.4.1
> (http://www.mpimet.mpg.de/cdo)
> dimensions:
> lon = 144
> lat = 73
> level = 17
> time = 1460 // unlimited
> variables:
> float lon ( lon )
> long_name : Longitude
> units : degrees_east
> standard_name : longitude
>
> float lat ( lat )
> long_name : Latitude
> units : degrees_north
> standard_name : latitude
>
> float level ( level )
> long_name : Level
> units : millibar
> axis : z
>
> double time ( time )
> units : hours since 1800-01-01 00:00:00
> calendar : standard
>
> float WSpeed ( time, level, lat, lon )
> long_name : Wind Speed
> standard_name : eastward_wind
> units : m/s
> _FillValue : -32767
> valid_range : ( -140, 174.989990234375 )
> parent_stat : Other
> statistic : Individual Obs
> level_desc : Pressure Levels
> dataset : NCEP/DOE AMIP-II Reanalysis (Reanalysis-2)
> var_desc : u-wind
> GRIB_name : UGRD
> GRIB_id : 33
> least_significant_digit : 1
> precision : 2
> actual_range : ( -87.30000305175781, 129.5 )
> unpacked_valid_range : ( -140, 175 )
>
>
>
>
> Thank you
>
> Vishali
>
>
>
>
>
> On Mon, Jun 20, 2011 at 4:06 PM, Dennis Shea <shea@ucar.edu
> <mailto:shea@ucar.edu>> wrote:
>
> You must help with more information:
>
> ==============================__====
> Try some different approaches. Isolate the following,
> as a separate script. run it
>
>
>
> diru = "/home/Vishali/Documents/__dadosVentos/Speed/"
> filu = systemfunc("cd "+diru+" ; ls wind*nc")
> nfilu = dimsizes(filu)
>
> print(nfilu)
> print(filu)
>
> ==============================__==
> Another separate script:
>
>
> yrStrt = 1998
> yrLast = 2004
> nyrs = yrLast-yrStrt+1
> klev = 17
> nlat = 73
> mlon = 144
> ntim = 12*nyrs ; number of months
>
>
> wMon = new ( (/ntim,klev,nlat,mlon/), "float", 1e20)
> tMon = new ( ntim, "double", "No_FillValue")
>
> print("success")
> ==============================
>
> Also, pick out (say) the 1st file and use ncl_filedump
>
> %> ncl_filedump wind...nc
>
>
> On 6/20/11 9:45 AM, Vishali P. wrote:
>
> Hi.
>
> When I tried to run normally, it only appeared : ' Segmentation
> fault
> (core dumped)'
>
> When I've tried to run with Mary's command ncl -x
> CalculateSeasonalMeans.ncl :
>
> Loading file
> "$NCARG_ROOT/lib/ncarg/__nclscripts/csm/contributed.__ncl"
> +
> +
> + yrStrt = 1998
> + yrLast = 2004
> + nyrs = yrLast-yrStrt+1
> + klev = 17
> + nlat = 73
> + mlon = 144
> + ntim = 12*nyrs ; number of months
> +
> + wMon = new ( (/ntim,klev,nlat,mlon/), "float", 1e20)
> + tMon = new ( ntim, "double", "No_FillValue")
> +
> +
> + diru = "/home/Vishali/Documents/__dadosVentos/SpeedTest/"
> + filu = systemfunc("cd "+diru+" ; ls wind*nc")
> + nfilu = dimsizes(filu)
> +
> + ; calculate monthly means from 6hrly
> + nmoStrt = 0
> + nmoLast = 11
> + do nf=0,nfilu-1
> +
> + f = addfile(diru+filu(nf), "r")
> + w = f->WSpeed
> + wMon(nmoStrt:nmoLast,:,:,:) = calculate_monthly_values(w,
> "avg",0, False)
> + nmoStrt = nmoStrt+12
> + nmoLast = nmoLast+12
> + delete(w) ; may change size next iteration (leap yr)
> + end do
> Segmentation fault (core dumped)
>
>
>
>
> On Mon, Jun 20, 2011 at 3:16 PM, Mary Haley <haley@ucar.edu
> <mailto:haley@ucar.edu>
> <mailto:haley@ucar.edu <mailto:haley@ucar.edu>>> wrote:
>
>
> On Jun 20, 2011, at 2:40 AM, Vishali P. wrote:
>
> > Dear NCL users
> >
> > I'm using 4x Daily NCEP/DOE AMIP-II Reanalysis wind data
> 1998 to
> 2004. I've calculated wind speed for each year...and created
> outputs.
> > I was suggested a script to calculate unweighted seasonal
> means.
> And then, average of each season from 1998 to 2004, ie,
> average if
> season 1 from 1998 to 2004,average of season 2 from 1998 to 2004,
> then 3rd and 4th.
> >
> > But when I try to run,appears error "Segmentation fault (core
> dumped)".
> > Can anyone help me ?
> > Current version of my NCL is 5.2.1.
> >
> >
> > Thank you in advance
> >
> > Vishali
>
> Hi Vishali,
>
> It could be a memory issue, but I don't have enough
> information to
> determine this.
>
> Do you know where the code is seg faulting? To get this
> information,
> run ncl with the -x option, which causes every line to be
> executed
> as it is run:
>
> ncl -x your_script.ncl
>
> It would help if you could include all the output from
> running this
> command, plus all the output from your various printVarSummary
> statements.
>
> Thanks,
>
> --Mary
>
> >
> >
>
> ==============================__==============================__=============================
> > load
> "$NCARG_ROOT/lib/ncarg/__nclscripts/csm/contributed.__ncl"
> >
> > yrStrt = 1998
> > yrLast = 2004
> > nyrs = yrLast-yrStrt+1
> > klev = 17
> > nlat = 73
> > mlon = 144
> > ntim = 12*nyrs ; number of months
> >
> > wMon = new ( (/ntim,klev,nlat,mlon/), "float", 1e20)
> > tMon = new ( ntim, "double", "No_FillValue")
> >
> >
> > diru = "/home/Vishali/Documents/__dadosVentos/Speed/"
> > filu = systemfunc("cd "+diru+" ; ls wind*nc")
> > nfilu = dimsizes(filu)
> >
> > ; calculate monthly means from 6hrly
> > nmoStrt = 0
> > nmoLast = 11
> > do nf=0,nfilu-1
> > f = addfile(diru+filu(nf), "r")
> > w = f->WSpeed
> > wMon(nmoStrt:nmoLast,:,:,:) =
> calculate_monthly_values(w,
> "avg",0, False)
> > nmoStrt = nmoStrt+12
> > nmoLast = nmoLast+12
> > delete(w) ; may change size next iteration (leap yr)
> > end do
> > delete(wMon&time)
> >
> > printVarSummary(wMon) ; wMon(84,17,73,144)
> >
> > time = yyyymm_time(yrStrt, yrLast, "integer")
> > wMon&time = time
> >
> > wMon!1 = "level" ; minor bug in
> "calculate_monthly_values"
> > wMon&level = f->level
> >
> > printVarSummary(wMon)
> >
> > ; unweighted seasonal means
> >
> > wAllSea = wgt_runave_n_Wrap(wMon, 3, 0, 0)
> > printVarSummary(wAllSea)
> >
> > wSea = wAllSea(::3,:,:,:) ; (28,14,73,144)
> > printVarSummary(wSea)
> >
> > _________________________________________________
> > ncl-talk mailing list
> > List instructions, subscriber options, unsubscribe:
> > http://mailman.ucar.edu/__mailman/listinfo/ncl-talk
> <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
>
>
>
>
> _________________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/__mailman/listinfo/ncl-talk
> <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Mon Jun 20 12:01:50 2011