/[MITgcm]/manual/s_phys_pkgs/text/exch2.tex
ViewVC logotype

Diff of /manual/s_phys_pkgs/text/exch2.tex

Parent Directory Parent Directory | Revision Log Revision Log | View Revision Graph Revision Graph | View Patch Patch

revision 1.12 by afe, Tue Mar 16 21:52:15 2004 UTC revision 1.14 by afe, Wed Mar 17 21:44:02 2004 UTC
# Line 21  sphere topology configuration to allow m Line 21  sphere topology configuration to allow m
21  decomposition and parallelization.  Cube faces (also called  decomposition and parallelization.  Cube faces (also called
22  subdomains) may be divided into any number of tiles that divide evenly  subdomains) may be divided into any number of tiles that divide evenly
23  into the grid point dimensions of the subdomain.  Furthermore, the  into the grid point dimensions of the subdomain.  Furthermore, the
24  individual tiles may be run on separate processors in different  individual tiles can run on separate processors in different
25  combinations, and whether exchanges between particular tiles occur  combinations, and whether exchanges between particular tiles occur
26  between different processors is determined at runtime.  This  between different processors is determined at runtime.  This
27  flexibility provides for manual compile-time load balancing across a  flexibility provides for manual compile-time load balancing across a
# Line 65  $\bullet$ An example of \file{W2\_EXCH2\ Line 65  $\bullet$ An example of \file{W2\_EXCH2\
65    configurations other than the one you intend to modify.\\    configurations other than the one you intend to modify.\\
66    
67  $\bullet$ Files containing grid parameters, named  $\bullet$ Files containing grid parameters, named
68    \file{tile00$n$.mitgrid} where $n$=[1,6] (one per subdomain), must    \file{tile00$n$.mitgrid} where $n$=\code{(1:6)} (one per subdomain),
69    be in the working directory when the MITgcm executable is run.    must be in the working directory when the MITgcm executable is run.
70    These files are provided in the example experiments for cubed sphere    These files are provided in the example experiments for cubed sphere
71    configurations with 32$\times$32 cube sides and are non-trivial to    configurations with 32$\times$32 cube sides and are non-trivial to
72    generate -- please contact MITgcm support if you want to generate    generate -- please contact MITgcm support if you want to generate
73    files for other configurations. \\    files for other configurations. \\
74    
75  $\bullet$ As always when compiling MITgcm, the file \file{SIZE.h} must  $\bullet$ As always when compiling MITgcm, the file \file{SIZE.h} must
76    be placed where \file{genmake2} will find it.  In particular for the    be placed where \file{genmake2} will find it.  In particular for
77    exch2, the domain decomposition specified in \file{SIZE.h} must    exch2, the domain decomposition specified in \file{SIZE.h} must
78    correspond with the particular configuration's topology specified in    correspond with the particular configuration's topology specified in
79    \file{W2\_EXCH2\_TOPOLOGY.h} and \file{w2\_e2setup.F}.  Domain    \file{W2\_EXCH2\_TOPOLOGY.h} and \file{w2\_e2setup.F}.  Domain
# Line 119  generated configuration are \code{nr}, \ Line 119  generated configuration are \code{nr}, \
119  The first three determine the size of the subdomains and  The first three determine the size of the subdomains and
120  hence the size of the overall domain.  Each one determines the number  hence the size of the overall domain.  Each one determines the number
121  of grid points, and therefore the resolution, along the subdomain  of grid points, and therefore the resolution, along the subdomain
122  sides in a ``great circle'' around each axis of the cube.  At the time  sides in a ``great circle'' around an axis of the cube.  At the time
123  of this writing MITgcm requires these three parameters to be equal,  of this writing MITgcm requires these three parameters to be equal,
124  but they provide for future releases  to accomodate different  but they provide for future releases  to accomodate different
125  resolutions around the axes to allow (for example) greater resolution  resolutions around the axes to allow (for example) greater resolution
# Line 129  The parameters \code{tnx} and \code{tny} Line 129  The parameters \code{tnx} and \code{tny}
129  the tiles into which the subdomains are decomposed, and must evenly  the tiles into which the subdomains are decomposed, and must evenly
130  divide the integer assigned to \code{nr}, \code{nb} and \code{ng}.  divide the integer assigned to \code{nr}, \code{nb} and \code{ng}.
131  The result is a rectangular tiling of the subdomain.  Figure  The result is a rectangular tiling of the subdomain.  Figure
132  \ref{fig:24tile} shows one possible topology for a twenty-four tile  \ref{fig:24tile} shows one possible topology for a twentyfour-tile
133  cube, and figure \ref{fig:12tile} shows one for twelve tiles. \\  cube, and figure \ref{fig:12tile} shows one for twelve tiles. \\
134    
135  \begin{figure}  \begin{figure}
# Line 139  cube, and figure \ref{fig:12tile} shows Line 139  cube, and figure \ref{fig:12tile} shows
139   }   }
140  \end{center}  \end{center}
141    
142  \caption{Plot of cubed sphere topology with a 32$\times$192 domain  \caption{Plot of a cubed sphere topology with a 32$\times$192 domain
143  divided into six 32$\times$32 subdomains, each of which is divided into four tiles  divided into six 32$\times$32 subdomains, each of which is divided into four tiles
144  (\code{tnx=16, tny=16}) for a total of twenty-four tiles.  (\code{tnx=16, tny=16}) for a total of twentyfour tiles.
145  } \label{fig:24tile}  } \label{fig:24tile}
146  \end{figure}  \end{figure}
147    
# Line 151  divided into six 32$\times$32 subdomains Line 151  divided into six 32$\times$32 subdomains
151    \includegraphics{part6/s12t_16x32.ps}    \includegraphics{part6/s12t_16x32.ps}
152   }   }
153  \end{center}  \end{center}
154  \caption{Plot of cubed sphere topology with a 32$\times$192 domain  \caption{Plot of a cubed sphere topology with a 32$\times$192 domain
155  divided into six 32$\times$32 subdomains of two tiles each  divided into six 32$\times$32 subdomains of two tiles each
156   (\code{tnx=16, tny=32}).   (\code{tnx=16, tny=32}).
157  } \label{fig:12tile}  } \label{fig:12tile}
158  \end{figure}  \end{figure}
159    
160    \begin{figure}
161    \begin{center}
162     \resizebox{4in}{!}{
163      \includegraphics{part6/s6t_32x32.ps}
164     }
165    \end{center}
166    \caption{Plot of a cubed sphere topology with a 32$\times$192 domain
167    divided into six 32$\times$32 subdomains with one tile each
168    (\code{tnx=32, tny=32}).  This is the default configuration.
169      }
170    \label{fig:6tile}
171    \end{figure}
172    
173    
174  Tiles can be selected from the topology to be omitted from being  Tiles can be selected from the topology to be omitted from being
175  allocated memory and processors.  This tuning is useful in ocean  allocated memory and processors.  This tuning is useful in ocean
176  modeling for omitting tiles that fall entirely on land.  The tiles  modeling for omitting tiles that fall entirely on land.  The tiles
# Line 171  by their tile number in the topology, se Line 185  by their tile number in the topology, se
185  \label{sec:exch2mpi}  \label{sec:exch2mpi}
186    
187  Once the topology configuration files are created, the Fortran  Once the topology configuration files are created, the Fortran
188  parameters in \file{SIZE.h} must be configured to match.  Section  \code{PARAMETER}s in \file{SIZE.h} must be configured to match.
189  \ref{sect:specifying_a_decomposition} \sectiontitle{Specifying a  Section \ref{sect:specifying_a_decomposition} \sectiontitle{Specifying
190  decomposition} provides a general description of domain decomposition  a decomposition} provides a general description of domain
191  within MITgcm and its relation to \file{SIZE.h}. The current section  decomposition within MITgcm and its relation to \file{SIZE.h}. The
192  specifies certain constraints the exch2 package imposes as well as  current section specifies certain constraints the exch2 package
193  describes how to enable parallel execution with MPI. \\  imposes as well as describes how to enable parallel execution with
194    MPI. \\
195    
196  As in the general case, the parameters \varlink{sNx}{sNx} and  As in the general case, the parameters \varlink{sNx}{sNx} and
197  \varlink{sNy}{sNy} define the size of the individual tiles, and so  \varlink{sNy}{sNy} define the size of the individual tiles, and so
# Line 266  generated by \file{driver.m}.\\ Line 281  generated by \file{driver.m}.\\
281  The scalar parameters \varlink{exch2\_domain\_nxt}{exch2_domain_nxt}  The scalar parameters \varlink{exch2\_domain\_nxt}{exch2_domain_nxt}
282  and \varlink{exch2\_domain\_nyt}{exch2_domain_nyt} express the number  and \varlink{exch2\_domain\_nyt}{exch2_domain_nyt} express the number
283  of tiles in the $x$ and $y$ global indices.  For example, the default  of tiles in the $x$ and $y$ global indices.  For example, the default
284  setup of six tiles has \code{exch2\_domain\_nxt=6} and  setup of six tiles (Fig. \ref{fig:6tile}) has \code{exch2\_domain\_nxt=6} and
285  \code{exch2\_domain\_nyt=1}.  A topology of twenty-four square tiles,  \code{exch2\_domain\_nyt=1}.  A topology of twenty-four square tiles,
286  four per subdomain (as in figure \ref{fig:24tile}), will have  four per subdomain (as in figure \ref{fig:24tile}), will have
287  \code{exch2\_domain\_nxt=12} and \code{exch2\_domain\_nyt=2}.  Note  \code{exch2\_domain\_nxt=12} and \code{exch2\_domain\_nyt=2}.  Note
288  that these parameters express the tile layout to allow global data  that these parameters express the tile layout to allow global data
289  files that are tile-layout-neutral and have no bearing on the internal  files that are tile-layout-neutral and have no bearing on the internal
290  storage of the arrays.  The tiles are internally stored in a range  storage of the arrays.  The tiles are internally stored in a range
291  from [1,\varlink{bi}{bi}] the $x$ axis and $y$ axis variable  from \code{(1:\varlink{bi}{bi})} the $x$ axis, and $y$ axis variable
292  \varlink{bj}{bj} is generally ignored within the package. \\  \varlink{bj}{bj} is generally ignored within the package. \\
293    
294  \subsubsection{Arrays Indexed to Tile Number}  \subsubsection{Arrays Indexed to Tile Number}
295    
296  The following arrays are of size \code{NTILES}, are indexed to the  The following arrays are of length \code{NTILES}, are indexed to the
297  tile number, and the indices are omitted in their descriptions. \\  tile number, and the indices are omitted in their descriptions. \\
298    
299  The arrays \varlink{exch2\_tnx}{exch2_tnx} and  The arrays \varlink{exch2\_tnx}{exch2_tnx} and
# Line 293  The location of the tiles' Cartesian ori Line 308  The location of the tiles' Cartesian ori
308  determined by the arrays \varlink{exch2\_tbasex}{exch2_tbasex} and  determined by the arrays \varlink{exch2\_tbasex}{exch2_tbasex} and
309  \varlink{exch2\_tbasey}{exch2_tbasey}.  These variables are used to  \varlink{exch2\_tbasey}{exch2_tbasey}.  These variables are used to
310  relate the location of the edges of different tiles to each other.  As  relate the location of the edges of different tiles to each other.  As
311  an example, in the default six-tile topology ??  each index in these  an example, in the default six-tile topology (Fig. \ref{fig:6tile})
312  arrays are set to \code{0}.  The twentyfour-tile case discussed above  each index in these arrays is set to \code{0} since a tile occupies
313  will have values of \code{0} or \code{16}, depending on the quadrant  its entire subdomain.  The twentyfour-tile case discussed above will
314  the tile falls within the subdomain.  The array  have values of \code{0} or \code{16}, depending on the quadrant the
315  \varlink{exch2\_myFace}{exch2_myFace} contains the number of the  tile falls within the subdomain.  The elements of the arrays
316  subdomain of each tile, numbered \code{(1:6)} in the case of the  \varlink{exch2\_txglobalo}{exch2_txglobalo} and
317  standard cube topology and indicated by \textbf{\textsf{f}}$n$ in  \varlink{exch2\_txglobalo}{exch2_txglobalo} are similar to
 figures \ref{fig:12tile}) and \ref{fig:24tile}). \\  
   
 The elements of the arrays \varlink{exch2\_txglobalo}{exch2_txglobalo}  
 and \varlink{exch2\_txglobalo}{exch2_txglobalo} are similar to  
318  \varlink{exch2\_tbasex}{exch2_tbasex} and  \varlink{exch2\_tbasex}{exch2_tbasex} and
319  \varlink{exch2\_tbasey}{exch2_tbasey}, but locate the tiles within the  \varlink{exch2\_tbasey}{exch2_tbasey}, but locate the tiles within the
320  global address space, similar to that used by global files. \\  global address space, similar to that used by global files. \\
321    
322    The array \varlink{exch2\_myFace}{exch2_myFace} contains the number of
323    the subdomain of each tile, in a range \code{(1:6)} in the case of the
324    standard cube topology and indicated by \textbf{\textsf{f}}$n$ in
325    figures \ref{fig:12tile} and
326    \ref{fig:24tile}. \varlink{exch2\_nNeighbours}{exch2_nNeighbours}
327    contains a count of how many neighboring tiles each tile has, and is
328    used for setting bounds for looping over neighboring tiles.
329    \varlink{exch2\_tProc}{exch2_tProc} holds the process rank of each
330    tile, and is used in interprocess communication.  \\
331    
332    
333  The arrays \varlink{exch2\_isWedge}{exch2_isWedge},  The arrays \varlink{exch2\_isWedge}{exch2_isWedge},
334  \varlink{exch2\_isEedge}{exch2_isEedge},  \varlink{exch2\_isEedge}{exch2_isEedge},
335  \varlink{exch2\_isSedge}{exch2_isSedge}, and  \varlink{exch2\_isSedge}{exch2_isSedge}, and
# Line 317  values are used within the topology gene Line 339  values are used within the topology gene
339  orientation of neighboring tiles, and to indicate whether a tile lies  orientation of neighboring tiles, and to indicate whether a tile lies
340  on the corner of a subdomain.  The latter case requires special  on the corner of a subdomain.  The latter case requires special
341  exchange and numerical handling for the singularities at the eight  exchange and numerical handling for the singularities at the eight
342  corners of the cube.  \varlink{exch2\_nNeighbours}{exch2_nNeighbours}  corners of the cube. \\
343  contains a count of how many neighboring tiles each tile has, and is  
 used for setting bounds for looping over neighboring tiles.  
 \varlink{exch2\_tProc}{exch2_tProc} holds the process rank of each  
 tile, and is used in interprocess communication.  \\  
344    
345  \subsubsection{Arrays Indexed to Tile Number and Neighbor}  \subsubsection{Arrays Indexed to Tile Number and Neighbor}
346    
# Line 336  in the order right to left on the north Line 355  in the order right to left on the north
355  to bottom on the east and west edges.  Maybe throw in a fig here, eh?  to bottom on the east and west edges.  Maybe throw in a fig here, eh?
356  \\  \\
357    
358    \sloppy
359  The \code{exch2\_opposingSend\_record(a,T)} array holds the index  The \code{exch2\_opposingSend\_record(a,T)} array holds the index
360  \code{b} in \texttt{exch2\_neighbourId(b,Tn)} that holds the tile  \code{b} in \texttt{exch2\_neighbourId(b,Tn)} that holds the tile
361  number \code{T}.  In other words,  number \code{T}.  In other words,
# Line 345  number \code{T}.  In other words, Line 365  number \code{T}.  In other words,
365  \end{verbatim}  \end{verbatim}
366  This provides a back-reference from the neighbor tiles. \\  This provides a back-reference from the neighbor tiles. \\
367    
368  The arrays \varlink{exch2\_pi}{exch2_pi},  The arrays \varlink{exch2\_pi}{exch2_pi} and
369  \varlink{exch2\_pj}{exch2_pj}, \varlink{exch2\_oi}{exch2_oi},  \varlink{exch2\_pj}{exch2_pj} specify the transformations of variables
370  \varlink{exch2\_oj}{exch2_oj}, \varlink{exch2\_oi\_f}{exch2_oi_f}, and  in exchanges between the neighboring tiles.  These transformations are
371  \varlink{exch2\_oj\_f}{exch2_oj_f} specify the transformations in  necessary in exchanges between subdomains because a physical vector
372  exchanges between the neighboring tiles.  The dimensions of  component in one direction may map to one in a different direction in
373  \code{exch2\_pi(t,N,T)} and \code{exch2\_pj(t,N,T)} are the neighbor  an adjacent subdomain, and may be have its indexing reversed. This
374  ID \code{N} and the tile number \code{T} as explained above, plus a  swapping arises from the ``folding'' of two-dimensional arrays into a
375  vector of length 2 containing transformation factors \code{t}.  The  three-dimensional cube.
376  first element of the transformation vector indicates the factor  
377  \code{t} by which variables representing the same vector component of  The dimensions of \code{exch2\_pi(t,N,T)} and \code{exch2\_pj(t,N,T)}
378  a tile \code{T} will be multiplied in exchanges with neighbor  are the neighbor ID \code{N} and the tile number \code{T} as explained
379  \code{N}, and the second element indicates the transform to the  above, plus a vector of length 2 containing transformation factors
380  variable in the other direction.  As an example,  \code{t}.  The first element of the transformation vector indicates
381  \code{exch2\_pi(1,N,T)} holds the transform of the $i$ component of a  the factor \code{t} by which variables representing the same
382  vector variable in tile \code{T} to the $i$ component of tile  \emph{physical} vector component of a tile \code{T} will be multiplied
383  \code{T}'s neighbor \code{N}, and \code{exch2\_pi(2,N,T)} hold the  in exchanges with neighbor \code{N}, and the second element indicates
384  component of neighbor \code{N}'s $j$ component. \\  the transform to the physical vector in the other direction.  To
385    clarify (hopefully), \code{exch2\_pi(1,N,T)} holds the transform of
386    the $i$ component of a vector variable in tile \code{T} to the $i$
387    component of tile \code{T}'s neighbor \code{N}, and
388    \code{exch2\_pi(2,N,T)} holds the transform of \code{T}'s $i$
389    components to the neighbor \code{N}'s $j$ component. \\
390    
391  Under the current cube topology, one of the two elements of  Under the current cube topology, one of the two elements of
392  \code{exch2\_pi} or \code{exch2\_pj} for a given tile \code{T} and  \code{exch2\_pi} or \code{exch2\_pj} for a given tile \code{T} and
393  neighbor \code{N} will be \code{0}, reflecting the fact that the two  neighbor \code{N} will be \code{0}, reflecting the fact that the two
394  vector components are orthogonal.  The other element will be 1 or -1,  vector components are orthogonal.  The other element will be \code{1}
395  depending on whether the components are indexed in the same or  or \code{-1}, depending on whether the components are indexed in the
396  opposite directions.  For example, the transform vector of the arrays  same or opposite directions.  For example, the transform vector of the
397  for all tile neighbors on the same subdomain will be \code{(1,0)},  arrays for all tile neighbors on the same subdomain will be
398  since all tiles on the same subdomain are oriented identically.  A  \code{(1,0)}, since all tiles on the same subdomain are oriented
399  vector direction that corresponds to the orthogonal dimension with the  identically.  A vector direction that corresponds to the orthogonal
400  same index direction in a particular tile-neighbor orientation will  dimension with the same index direction in a particular tile-neighbor
401  have \code{(0,1)}, whereas those in the opposite index direction will  orientation will have \code{(0,1)}, whereas those in the opposite
402  have \code{(0,-1)}.  This needs some diagrams.  index direction will have \code{(0,-1)}. \\
403    
404    The arrays \varlink{exch2\_oi}{exch2_oi},
405    \varlink{exch2\_oj}{exch2_oj}, \varlink{exch2\_oi\_f}{exch2_oi_f}, and
406    \varlink{exch2\_oj\_f}{exch2_oj_f} are indexed to tile number and
407    neighbor and specify the relative offset within the subdomain of the
408    array index of a variable going from a neighboring tile $N$ to a local
409    tile $T$.  Consider the six-tile case (Fig. \ref{fig:6tile}), where
410    \code{exch2\_oi(1,1)=33}, \code{exch2\_oi(2,1)=0},
411    \code{exch2\_oi(3,1)=32}, and \code{exch2\_oi(4,1)=-32}.  Each of these
412    indicates the offset in the $x$ direction \\
413    
414    Finally, \varlink{exch2\_itlo\_c}{exch2_itlo_c},
415    \varlink{exch2\_ithi\_c}{exch2_ithi_c},
416    \varlink{exch2\_jtlo\_c}{exch2_jtlo_c} and
417    \varlink{exch2\_jthi\_c}{exch2_jthi_c} hold the location and index
418    bounds of the edge segment of the neighbor tile \code{N}'s subdomain
419    that gets exchanged with the local tile \code{T}.  To take the example
420    of tile \code{T=2} in the twelve-tile topology
421    (Fig. \ref{fig:12tile}): \\
422    
423    \begin{verbatim}
424           exch2_itlo_c(4,2)=17
425           exch2_ithi_c(4,2)=17
426           exch2_jtlo_c(4,2)=0
427           exch2_jthi_c(4,2)=33
428    \end{verbatim}
429    
430    Here \code{N=4}, indicating the western neighbor, which is \code{Tn=1}.
431    \code{Tn=1} resides on the same subdomain as \code{T=2}, so the tiles
432    have the same orientation and the same $x$ and $y$ axes.  The $i$
433    component is orthogonal to the western edge and the tile is 16 points
434    wide, so \code{exch2\_itlo\_c} and \code{exch2\_ithi\_c} indicate the
435    column beyond \code{Tn=1}'s eastern edge, in that tile's halo
436    region. Since the border of the tiles extends through the entire
437    height of the subdomain, the $y$ axis bounds \code{exch2\_jtlo\_c} to
438    \code{exch2\_jthi\_c} cover the height, plus 1 in either direction to
439    cover part of the halo. \\
440    
441    For the north edge of the same tile \code{T=2} where \code{N=1} and
442    the neighbor tile is \code{Tn=5}:
443    
 {\footnotesize  
444  \begin{verbatim}  \begin{verbatim}
445  C      exch2_pi          :: X index row of target to source permutation         exch2_itlo_c(1,2)=0
446  C                        :: matrix for each neighbour entry.                     exch2_ithi_c(1,2)=0
447  C      exch2_pj          :: Y index row of target to source permutation         exch2_jtlo_c(1,2)=0
448  C                        :: matrix for each neighbour entry.                     exch2_jthi_c(1,2)=17
 C      exch2_oi          :: X index element of target to source  
 C                        :: offset vector for cell-centered quantities    
 C                        :: of each neighbor entry.                      
 C      exch2_oj          :: Y index element of target to source  
 C                        :: offset vector for cell-centered quantities    
 C                        :: of each neighbor entry.                      
 C      exch2_oi_f        :: X index element of target to source  
 C                        :: offset vector for face quantities            
 C                        :: of each neighbor entry.                      
 C      exch2_oj_f        :: Y index element of target to source  
 C                        :: offset vector for face quantities            
 C                        :: of each neighbor entry.                      
449  \end{verbatim}  \end{verbatim}
450  }  
451    \code{T}'s northern edge is parallel to the $x$ axis, but since
452    \code{Tn}'s $y$ axis corresponds to \code{T}'s $x$ axis,
453    \code{T}'s northern edge exchanges with \code{Tn}'s western edge.
454    The western edge of the tiles corresponds to the lower bound of the
455    $x$ axis, so \code{exch2\_itlo\_c} \code{exch2\_ithi\_c} are \code{0}. The
456    range of \code{exch2\_jtlo\_c} and \code{exch2\_jthi\_c} correspond to the
457    width of \code{T}'s northern edge, plus the halo. \\
458    
459    
460    
461    
462    
463    
464    
465    
466    
467    
468    
469    This needs some diagrams. \\
470    
471    
472    

Legend:
Removed from v.1.12  
changed lines
  Added in v.1.14

  ViewVC Help
Powered by ViewVC 1.1.22