1dgmap(1)                    PT-Scotch user's manual                   dgmap(1)
2
3
4

NAME

6       dgmap, dgpart - compute static mappings and partitions in parallel
7

SYNOPSIS

9       dgmap [options] [gfile] [tfile] [mfile] [lfile]
10
11       dgpart [options] [nparts] [gfile] [mfile] [lfile]
12

DESCRIPTION

14       The  dgmap  program  computes, in a parallel way, a static mapping of a
15       source graph onto a target graph.
16
17       The dgpart program is a shortcut of dgmap for computing partitions of a
18       source graph.
19
20       Source graph file gfile is either a centralized graph file, or a set of
21       files representing fragments of a distributed  graph.  For  dgmap,  the
22       target  architecture  file tfile describes either algorithmically-coded
23       topologies such as  meshes  and  hypercubes,  or  decomposition-defined
24       architectures  created  by means of the amk_grf(1) program. See gmap(1)
25       for a description of target architectures.  The  resulting  mapping  is
26       stored  in  file  mfile.  Eventual logging information (such as the one
27       produced by option -v) is sent to file lfile. When file names  are  not
28       specified,  data  is  read  from standard input and written to standard
29       output. Standard streams can also be explicitely represented by a  dash
30       '-'.
31
32       When the proper libraries have been included at compile time, dgmap and
33       dgpart can directly handle compressed graphs, both as input and output.
34       A stream is treated as compressed whenever its name is postfixed with a
35       compressed file extension, such as in supported are  the  bzip2  format
36       ('.bz2'),  the  gzip  format  ('.gz'), and the lzma format ('.lzma', on
37       input only).
38
39       dgmap and dgpart base on implementations of the MPI interface to spread
40       work  across  the processing elements. They are therefore not likely to
41       be run directly, but instead through  some  launcher  command  such  as
42       mpirun.
43

OPTIONS

45       -h     Display some help.
46
47       -mstrat
48              Use parallel mapping strategy strat (see PT-Scotch user's manual
49              for more information).
50
51       -rpnum Set root process for centralized files (default is 0).
52
53       -V     Display program version and copyright.
54
55       -vverb Set verbose mode to verb. It is a set of one of more  characters
56              which can be:
57
58              m      mapping information.
59
60              s      strategy information.
61
62              t      timing information.
63

NOTE

65       At  the  time  being  (version 5.1.0), dgmap cannot compute full static
66       mappings as gmap(1) does, but only partitions (that is,  mappings  onto
67       unweighted  or  weighted  complete  graphs). Target architectures other
68       than the 'cmplt' and 'wcmplt' ones will lead to an error message.
69

EXAMPLES

71       Run dgpart on 5 processing elements to compute a partition into 7 parts
72       of graph brol.grf and save the resulting ordering to file brol.map.
73
74             $ mpirun -np 5 dgpart 7 brol.grf brol.map
75
76       Run  dgpart on 5 processing elements to partition into 7 parts the dis‐
77       tributed  graph  stored  on  graph  fragment   files   brol5-0.dgr   to
78       brol5-4.dgr,  and  save  the  resulting  mapping  to file brol.map (see
79       dgscat(1) for an explanation of the '%p' and '%r' sequences in names of
80       distributed graph fragments).
81
82             $ mpirun -np 5 dgpart 7 brol%p-%r.dgr brol.map
83
84

SEE ALSO

86       dgtst(1), dgscat(1), amk_grf(1), acpl(1), gmap(1), gmtst(1).
87
88       PT-Scotch user's manual.
89

AUTHOR

91       Francois Pellegrini <francois.pellegrini@labri.fr>
92
93
94
95                              September 08, 2008                      dgmap(1)