[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20030909: 20030905: 20030904: GDDIAG Error



Patrick,

You need to ensure that gemprm.h uses the same size for LLMXGD to
ensure the C and Fortran routies are talking the same. LLMDGG is
a multiple of the largest grid that will be used in computations.

Realistically, you are limited to the size of your stack limits
of the system  for the executable (if the size is to big, you'll
segment fault at runtime).

The parameter LLMXTG is new, it is the largest grid
that can be stored, while LLMXGD is the largest subset
grid that can be manipulated. LLMDGG is the computational space
necessary for the number of grids involved in a computation.
More complicated computations that involve multiple grids will
use the most space.

The LLMXTG was added to allow you to decode the entire grid
such as the NDFD, and use the IJSKIP parameter and GAREA to
to subset the grid or reduce the number of points (eg, if you don't 
need 5km spacing when viewing on the CONUS scale).

Steve Chiswell




>From: "Patrick O'Reilly" <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200309091535.h89FZKLd009615

>Steve,
>
>When I set LLMXGD, is there an upper limit to what it can be set to?  From
>what I understand, I need to:
>
>Redefine LLMXGD and LLMDGG in $GEMPAK/include/gemprm.h and MCHPRM.xxxxx
>
>Remove all previously compiled libraries (make distclean from
>~gempak/GEMPAK5.6)
>
>rebuild the package completely (make all)
>
>Thanks for sharing your immense knowledge.
>
>Patrick
>
>----- Original Message ----- 
>From: "Unidata Support" <address@hidden>
>To: "Patrick O'Reilly" <address@hidden>
>Cc: "Unidata Support" <address@hidden>
>Sent: Sunday, September 07, 2003 4:13 PM
>Subject: 20030905: 20030904: GDDIAG Error
>
>
>>
>> Patrick,
>>
>> The internal grid space for your calculation (LLMDGG) must be large enough
>> to accomodate 1121x881xN grid points, where N is the number of parameters
>> involved in the calculation (eg, here it is 3 for C = MUL (a,b)). This
>would be
>> 2,962,803 for your grids. If you change LLMXGD, then you must also set
>LLMDGG accordingly
>> for your computations. The default of 2,500,000 must be increased to at
>least
>> 3,000,000.
>>
>> Steve Chiswell
>>
>>
>>
>>
>>
>> >From: "Patrick O'Reilly" <address@hidden>
>> >Organization: UCAR/Unidata
>> >Keywords: 200309051932.h85JWPLd024016
>>
>> >Here's the info on the file I am sending (assuming it makes it).
>> >
>> > GEMPAK-GDINFO>r
>> >
>> > GRID FILE: test.gem
>> >
>> > GRID NAVIGATION:
>> >     PROJECTION:          STR
>> >     ANGLES:                90.0   255.0     0.0
>> >     GRID SIZE:         1121 881
>> >     LL CORNER:              23.12   -119.02
>> >     UR CORNER:              45.62    -59.95
>> >
>> > GRID ANALYSIS BLOCK:
>> >     ANALYSIS TYPE:        BARNES
>> >     DELTAN:               4.000
>> >     DELTAX:           -9999.000
>> >     DELTAY:           -9999.000
>> >     GRID AREA:            19.00 -135.00   59.00  -59.00
>> >     EXTEND AREA:          19.00 -135.00   59.00  -59.00
>> >     DATA AREA:            19.00 -135.00   59.00  -59.00
>> >
>> > Number of grids in file:     2
>> >
>> > Maximum number of grids in file:    200
>> >
>> >  NUM       TIME1              TIME2           LEVL1 LEVL2  VCORD PARM
>> >    1     030731/1200F024   030830/1200F024        0         NONE AVEP24M
>> >    2     030731/1200F024   030830/1200F024        0         NONE CNTP24M
>> > Parameters requested: GDFILE,LSTALL,OUTPUT,GDATTIM,GLEVEL,GVCORD,GFUNC.
>> > GEMPAK-GDINFO>
>> >
>> >
>> >I am running GEMPAK version 5.6.h on Solaris 8.  Take note of the grid
>size,
>> >I had to re-build GEMPAK to run on larger grids than was the default.  I
>am
>> >getting these precip grids from
>> >ftp://ftpprd.ncep.noaa.gov/pub/gcp/precip/stage2/20030905/ and the file
>name
>> >is ST2un2003090512.24h.Z.  Thanks!
>> >
>> >
>> >Patrick
>> >
>> >----- Original Message ----- 
>> >From: "Unidata Support" <address@hidden>
>> >To: "Patrick O'Reilly" <address@hidden>
>> >Cc: <address@hidden>
>> >Sent: Friday, September 05, 2003 2:04 PM
>> >Subject: 20030904: GDDIAG Error
>> >
>> >
>> >>
>> >> Patrick,
>> >>
>> >> I ran a test on some grids that I had and it seems to work for
>> >> me like:
>> >>
>> >>  GDFILE   = test1.gem
>> >>  GDOUTF   = test1.gem
>> >>  GFUNC    = mul(avep24m,cntp24m)
>> >>  GDATTIM  = 030905/0000F024:030905/0600F024
>> >>  GLEVEL   = 0
>> >>  GVCORD   = none
>> >>  GRDNAM   = augp
>> >>  GPACK    =
>> >>  GEMPAK-GDDIAG>r
>> >>
>> >>
>> >>     TIME1             TIME2         LEVL1 LEVL2   VCORD PARM
>> >> 030905/0000F024    030905/0600F024      0          NONE AUGP
>> >> Enter a new grid parameter name, <cr> to accept or type EXIT:
>> >>  Parameters requested:
>GDFILE,GDOUTF,GFUNC,GDATTIM,GLEVEL,GVCORD,GRDNAM,
>> >>  GPACK.
>> >>  GEMPAK-GDDIAG>
>> >>
>> >> If you want to GDMOD the 2 grids in question to a file
>> >> and just send me that, I can check your grids here.
>> >>
>> >> Also let me know the OS and version you are using.
>> >>
>> >> It almost sounds like one of your grids is empty or corrupt. Have you
>> >> verified that you didn't hit the max number of grids in your file
>> >> along the way, etc? Of couse, if you can see them independently
>> >> with gdlist etc, then that isn't the problem.
>> >>
>> >> Steve Chiswell
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> >From: "Patrick O'Reilly" <address@hidden>
>> >> >Organization: UCAR/Unidata
>> >> >Keywords: 200309042022.h84KMbLd029630
>> >>
>> >> >Okay, what I'm trying to do is take a grid file with a months worth of
>> >daily
>> >> >24-hour radar estimated precip grids (P24M), and add them to get a
>> >monthly
>> >> >total radar estimated precip grid.
>> >> >
>> >> >So, I made a grid file using GDCFIL.  That was good.  Then I used
>GDSTAT
>> >to
>> >> >calculate the AVE, SDV, and CNT of the grids.  That went great.  Next,
>I
>> >> >tried using GDDIAG to multiply the AVEP24M and CNTP24M fields to get a
>> >grid
>> >> >with the total precip.  Not so good.  What I do get is:
>> >> >
>> >> > GEMPAK-GDDIAG>l
>> >> > GDFILE   = aug_precip.gem
>> >> > GDOUTF   = aug_precip.gem
>> >> > GFUNC    = MUL[AVEP24M,CNTP24M]
>> >> > GDATTIM  = 030731/1200f024:030830/1200f024
>> >> > GLEVEL   = 0
>> >> > GVCORD   = none
>> >> > GRDNAM   = augp
>> >> > GPACK    =
>> >> > GEMPAK-GDDIAG>r
>> >> > [DG -10]  Internal grid list is full; simplify function.
>> >> > [DG -10]  Internal grid list is full; simplify function.
>> >> > Parameters requested:
>GDFILE,GDOUTF,GFUNC,GDATTIM,GLEVEL,GVCORD,GRDNAM,
>> >> > GPACK.
>> >> > GEMPAK-GDDIAG>
>> >> >
>> >> >Couldn't find much info on the support archive list about this error,
>> >only
>> >> >that the GFUNC may be too long, but in this case it isn't.  Any
>pointers?
>> >> >
>> >> >Patrick
>> >> >_______________________________________
>> >> >Patrick O'Reilly
>> >> >Meteorological Decision Support Scientist
>> >> >The STORM Project - University of Northern Iowa
>> >> >address@hidden  ~  ph: 319-273-3789
>> >> >http://www.uni.edu/storm
>> >> >
>> >> >"No trees were killed in the making of this e-mail...however,
>> >> >a large number of electrons were horribly inconvenienced."
>> >> >
>> >>
>> >>
>>
>>***************************************************************************
>*
>> ><
>> >> Unidata User Support                                    UCAR Unidata
>> >> (303)497-8643                                                  P.O. Box
>> >> address@hidden                                   Boulder, CO
>>
>>> -------------------------------------------------------------------------
>-
>> >> Unidata WWW Service
>> >>
>>
>>***************************************************************************
>*
>> ><
>> >
>>
>>
>****************************************************************************
><
>> Unidata User Support                                    UCAR Unidata
>> (303)497-8643                                                  P.O. Box
>> address@hidden                                   Boulder, CO
>> --------------------------------------------------------------------------
>> Unidata WWW Service
>>
>****************************************************************************
><
>