[RegCNET] Error in Parallel Runing

[Maurice.McHugh at noaa.gov]

If you read the PGI comipiler's errors closely you will see that this is a compilation issue and not a parallel-programming/running or MPICH issue.

Please ensure that you're using the makefile appropriate to your system.

Good luck!

----- Original Message -----
From: serein05 <serein05 at 126.com>
Date: Thursday, March 25, 2010 5:30 am
Subject: [RegCNET] Error in Parallel Runing
To: regcnet <regcnet at lists.ictp.it>


> Dear all, 
> 
> While runing RegCM3 on a Linux Cluster(PGI 7.1 + MPICH), the error of 
> Makefile came out. 
> 
> PGF90-S-0087-Non-constant expression where constant expression 
> required (./parame: 39)
> PGF90-S-0087-Non-constant expression where constant expression 
> required (./parame: 39)
> PGF90-S-0087-Non-constant expression where constant expression 
> required (./parame: 39)
> PGF90-S-0087-Non-constant expression where constant expression 
> required (./Commons/parrad.cb: 21)
> PGF90-S-0087-Non-constant expression where constant expression 
> required (./Commons/parrad.cb: 47)
> PGF90-S-0050-Assumed size array, qvb3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, tb3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, ubx3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, vbx3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, ubd3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, vbd3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, rhob3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, qcb3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, rhb3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, qsb3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, pb3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, chib3d, is not a dummy argument 
> (./Common2/slice.cb: 5)
> PGF90-S-0050-Assumed size array, icumtop, is not a dummy argument 
> (./Common2/trachem.cb: 17)
> PGF90-S-0050-Assumed size array, icumbot, is not a dummy argument 
> (./Common2/trachem.cb: 17)
> PGF90-S-0050-Assumed size array, icumdwd, is not a dummy argument 
> (./Common2/trachem.cb: 17)
> PGF90-S-0050-Assumed size array, remrat, is not a dummy argument 
> (./Common2/trachem.cb: 24)
> PGF90-S-0050-Assumed size array, remlsc, is not a dummy argument 
> (./Common2/trachem.cb: 24)
> PGF90-S-0050-Assumed size array, remcvc, is not a dummy argument 
> (./Common2/trachem.cb: 24)
> PGF90-S-0050-Assumed size array, remdrd, is not a dummy argument 
> (./Common2/trachem.cb: 24)
> PGF90-S-0050-Assumed size array, rxsg, is not a dummy argument 
> (./Common2/trachem.cb: 24)
> PGF90-F-0008-Error limit exceeded (./Common2/trachem.cb: 24)
> PGF90/x86-64 Linux 7.1-4: compilation aborted
> make: *** [aermix.o] Error 2
> 
> 1. The parallel environment of my machine is fine, because the wrf 
> model ran successfully.
> 2. I followed the steps:
>     set NPROC>0
>     ln -s 0options/0_NODIAG_PARALLEL_CODE MAKECODE 
>     ./MAKECODE
> 3. The "Makefile" in Main directory is
>     
>    FFLAGS = -O1 -Ktrap=fp -byteswapio
>    MPI_ROOT = /cluster/server/mpich
>    # DEBUG --
>    # FFLAGS = -Mbounds -Mcache_align -Mchkfpstk -Mchkstk -Mdalign 
> -Mdclchk -Mdepchk -Miomutex -Mrecursive -Msave -Ktrap=fp -g -byteswapio
>    FC = $(MPI_ROOT)/bin/mpif90
>    CC = $(MPI_ROOT)/bin/mpicc
>    LIBS = -lm
> 
>    LDFLAG = -Bdynamic
>    LD = $(FC)
>    SRC = \
>    aermix.f ...
>    OBJ = \
>    aermix.o ...
>    all: regcm
>    regcm:  $(OBJ)
>            $(LD) -o $@ $(FFLAGS) $(OBJ)
>    clean:
>            rm -f $(OBJ) regcm *.o
> 
> 
> However, the problem mentioned above emerges when i run "mpiexec -n 4 
> ./regcm.x".
> 
> --
> With regard
> 
> 2010-03-25 
> 
> 
> 
> Matthew Chan
> _______________________________________________
> RegCNET mailing list
> RegCNET at lists.ictp.it