1gmap(1)                      Scotch user's manual                      gmap(1)
2
3
4

NAME

6       gmap, gpart - compute static mappings and partitions sequentially
7

SYNOPSIS

9       gmap [options] [gfile] [tfile] [mfile] [lfile]
10
11       gpart [options] [nparts] [gfile] [mfile] [lfile]
12

DESCRIPTION

14       The  gmap  program computes, in a sequential way, a static mapping of a
15       source graph onto a target graph.
16
17       The gpart program is a shortcut of gmap for computing unweighted parti‐
18       tions of a source graph.
19
20       Source graph file gfile can only be a centralized graph file. For gmap,
21       the target architecture file tfile  describes  either  algorithmically-
22       coded  topologies  such  as  meshes  and  hypercubes, or decomposition-
23       defined architectures created by means of the amk_grf(1)  program.  The
24       resulting mapping is stored in file mfile. Eventual logging information
25       (such as the one produced by option -v) is sent  to  file  lfile.  When
26       file  names  are  not  specified,  data is read from standard input and
27       written to standard output. Standard streams can  also  be  explicitely
28       represented by a dash '-'.
29
30       When  the proper libraries have been included at compile time, gmap and
31       gpart can directly handle compressed graphs, both as input and  output.
32       A stream is treated as compressed whenever its name is postfixed with a
33       compressed file extension, such as in supported are  the  bzip2  format
34       ('.bz2'),  the  gzip  format  ('.gz'), and the lzma format ('.lzma', on
35       input only).
36

OPTIONS

38       -h     Display some help.
39
40       -mstrat
41              Use sequential mapping strategy strat (see Scotch user's  manual
42              for more information).
43
44       -V     Display program version and copyright.
45
46       -vverb Set  verbose mode to verb. It is a set of one of more characters
47              which can be:
48
49              m      mapping information.
50
51              s      strategy information.
52
53              t      timing information.
54

TARGET ARCHITECTURES

56       Target architectures represent graphs  onto  which  source  graphs  are
57       mapped.  In  order  to  speed-up  the obtainment of target architecture
58       topological properties during the computation of mappings, some classi‐
59       cal  topologies are algorithmically coded into the mapper itself. These
60       topologies are consequently simply defined by their code name, followed
61       by their dimensional parameters:
62
63       cmplt dim
64              unweighted complete graph of size dim.
65
66       cmpltw dim w0 w1 ... wdim-1
67              weighted complete graph of size size and of respective loads w0,
68              w1, ..., wdim-1.
69
70       hcub dim
71              hypercube of dimension dim.
72
73       leaf hgt clu wgt
74              tree-leaf graph of height hgt with 2^hgt  vertices,  clu  levels
75              before  reaching  the  processing elements, and an inter-cluster
76              link weight of wgt.
77
78       mesh2D dimX dimY
79              2D mesh of dimX times dimY nodes.
80
81       mesh3D dimX dimY dimZ
82              23 mesh of dimX times dimY times dimZ nodes.
83
84       torus2D dimX dimY
85              2D torus of dimX times dimY nodes.
86
87       torus3D dimX dimY dimZ
88              3D torus of dimX times dimY times dimZ nodes.
89
90       Other target topologies can be created from their source graph descrip‐
91       tion by using the amk_grf(1) command. In this case, the target descrip‐
92       tion will begin with the code name deco.
93

MAPPINGS

95       Mappings are represented by as many lines as there are vertices in  the
96       source graph. Each of these lines is made of two figures: the number of
97       the vertex (or its label if source graph vertices are labeled) and  the
98       index of the target vertex to which it has been assigned. Target vertex
99       indices range from 0 to the number of vertices in the target  architec‐
100       ture (that is, the number of parts) minus one.
101
102       This  block of lines is always preceded by the number of such lines. In
103       most cases, since full mappings are requested, the number of  lines  is
104       equal to the number of vertices in the source graph.
105

EXAMPLES

107       Run  gpart  to  compute  a partition into 7 parts of graph brol.grf and
108       save the resulting ordering to file brol.map.
109
110             $ gpart 7 brol.grf brol.map
111
112       Run gmap to compute a partition, into 3 parts of respective weights  1,
113       2  and  4,  of  graph  brol.grf  and save the resulting mapping to file
114       brol.map. The dash '-' standard file name is used so  that  the  target
115       architecture  description  is read from the standard input, through the
116       pipe, as provided by the 'echo' shell command.
117
118            $ echo "cmpltw 3 1 2 4" | gmap brol.grf - brol.map
119
120

SEE ALSO

122       amk_grf(1), acpl(1), gmtst(1), dgmap(1).
123
124       Scotch user's manual.
125

AUTHOR

127       Francois Pellegrini <francois.pellegrini@labri.fr>
128
129
130
131                              September 08, 2008                       gmap(1)
Impressum