[MITgcm-support] 3d segmentation fault in MPI

m. r. schaferkotter schaferk at bellsouth.net
Sun Aug 29 15:44:22 EDT 2004


greetings:
i/ve successfully run the 2d lock exchange problem 1600x1x50 with 8 
processors (8 tiles; 200x50).

now i/d like to try the problem in 3d with 4 processors for testing.

a segmentation fault occurs from the start.

i include SIZE.h below
have i missed something?


       PARAMETER (
      &           sNx = 400,
      &           sNy = 200,
      &           OLx =   3,
      &           OLy =   3,
      &           nSx =   1,
      &           nSy =   1,
      &           nPx =   4,
      &           nPy =   1,
      &           Nx  = sNx*nSx*nPx,
      &           Ny  = sNy*nSy*nPy,
      &           Nr  =  50)

in data only one line is added
  delYfile='dy.bin',

in gendata

% Dimensions of grid
nx=1600;
ny=200;
nz =50;

the bathymetry file is converted into 2d. the bathymetry is a constant 
depth.
d=zeros(nx,ny);
for i=1:nx
   for j=1:ny
     d(i,j) = -50.0;
   end
end

fid=fopen('topog.slope','w',ieee); fwrite(fid,d,prec); fclose(fid);

in 2d T.init was created as:
t = zeros(nx,nz);
for i=1:nx
   for k=1:nz
     t(i,k) = (tmin - (dtmp/2.0)*(tanh((x(i) - xf)/scale)));
   end
end

in 3d.
t3d = zeros(nx,ny,nz);
for i=1:nx
   for j=1:ny
     for k=1:nz
       t3d(i,j,k) = (tmin - (dtmp/2.0)*(tanh((x(i) - xf)/scale)));
     end
   end
end

any ideas?

michael




More information about the MITgcm-support mailing list