1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
|
.\" Text automatically generated by txt2man
.TH dgmap 1 "23 November 2019" "" "PT-Scotch user's manual"
.SH NAME
\fBdgmap, dgpart \fP- compute static mappings and partitions in parallel
\fB
.SH SYNOPSIS
.nf
.fam C
\fBdgmap\fP [\fIoptions\fP] [\fIgfile\fP] [\fItfile\fP] [\fImfile\fP] [\fIlfile\fP]
\fBdgpart\fP [\fIoptions\fP] [\fInparts/\fIpwght\fP\fP] [\fIgfile\fP] [\fImfile\fP] [\fIlfile\fP]
.fam T
.fi
.fam T
.fi
.SH DESCRIPTION
The \fBdgmap\fP program computes, in a parallel way, a static mapping of a
source graph onto a target graph.
.PP
The \fBdgpart\fP program is a simplified interface to \fBdgmap\fP, which
performs graph partitioning instead of static mapping. Consequently,
the desired number of parts has to be provided, in lieu of the
target architecture. When using the program for graph clustering,
the number of parts turns into maximum cluster weight.
.PP
The \fB-b\fP and \fB-c\fP options allow the user to set preferences on the
behavior of the mapping strategy which is used by default. The \fB-m\fP
option allows the user to define a custom mapping strategy.
.PP
The \fB-q\fP option turns the programs into graph clustering programs. In
this case, \fBdgmap\fP only accepts variable-sized target architectures.
.PP
Source graph file \fIgfile\fP is either a centralized graph file, or a set
of files representing fragments of a distributed graph. For \fBdgmap\fP,
the target architecture file \fItfile\fP describes either algorithmically-coded
topologies such as meshes and hypercubes, or decomposition-defined
architectures created by means of the \fBamk_grf\fP(1) program. See
\fBgmap\fP(1) for a description of target architectures. The resulting
mapping is stored in file \fImfile\fP. Eventual logging information (such
as the one produced by option \fB-v\fP) is sent to file \fIlfile\fP. When file
names are not specified, data is read from standard input and
written to standard output. Standard streams can also be explicitely
represented by a dash '-'.
.PP
When the proper libraries have been included at compile time, \fBdgmap\fP
and \fBdgpart\fP can directly handle compressed graphs, both as input and
output. A stream is treated as compressed whenever its name is
postfixed with a compressed file extension, such as
in 'brol.grf.bz2' or '-.gz'. The compression formats which can be
supported are the bzip2 format ('.bz2'), the gzip format ('.gz'),
and the lzma format ('.lzma').
.PP
\fBdgmap\fP and \fBdgpart\fP base on implementations of the MPI interface to
spread work across the processing elements. They are therefore not
likely to be run directly, but instead through some launcher command
such as \fBmpirun\fP.
.SH OPTIONS
.TP
.B
\fB-b\fP\fIval\fP
Set maximum load imbalance ratio for graph partitioning
or static mapping. When programs are used as clustering
tools, this parameter sets the maximum load imbalance
ratio for recursive bipartitions. Exclusive with
the \fB-m\fP option.
.TP
.B
\fB-c\fP\fIopt\fP
Choose default mapping strategy according to one or
several options among:
.RS
.TP
.B
b
enforce load balance as much as possible.
.TP
.B
q
privilege quality over speed (default).
.TP
.B
s
privilege speed over quality.
.TP
.B
t
enforce safety.
.TP
.B
x
enforce scalability.
.PP
It is exclusive with the \fB-m\fP option.
.RE
.RE
.PP
.RS
.TP
.B
\fB-h\fP
Display some help.
.TP
.B
\fB-m\fP\fIstrat\fP
Use parallel mapping strategy \fIstrat\fP (see
PT-Scotch user's manual for more information).
.TP
.B
\fB-q\fP
(for \fBdgpart\fP)
.TP
.B
\fB-q\fP\fIpwght\fP
(for \fBdgmap\fP)
Use the programs as graph clustering tools instead of
static mapping or graph partitioning tools. For \fBdgpart\fP,
the number of parts will become the maximum cluster
weight. For \fBdgmap\fP, this number \fIpwght\fP has to be passed
after the option.
.TP
.B
\fB-r\fP\fIpnum\fP
Set root process for centralized files (default is 0).
.TP
.B
\fB-V\fP
Display program version and copyright.
.TP
.B
\fB-v\fP\fIverb\fP
Set verbose mode to \fIverb\fP. It is a set of one of more
characters which can be:
.RS
.TP
.B
m
mapping information.
.TP
.B
s
strategy information.
.TP
.B
t
timing information.
.SH NOTE
At the time being (version 5.1), \fBdgmap\fP cannot compute full static
mappings as \fBgmap\fP(1) does, but only partitions (that is, mappings
onto unweighted or weighted complete graphs). Target architectures
other than the 'cmplt' and 'wcmplt' ones will lead to an error
message.
.SH EXAMPLES
Run \fBdgpart\fP on 5 processing elements to compute a partition into 7
parts of graph brol.grf and save the resulting ordering to file brol.map.
.PP
.nf
.fam C
$ mpirun -np 5 dgpart 7 brol.grf brol.map
.fam T
.fi
Run \fBdgpart\fP on 5 processing elements to partition into 7 parts the
distributed graph stored on graph fragment files brol5-0.dgr to
brol5-4.dgr, and save the resulting mapping to file brol.map (see
\fBdgscat\fP(1) for an explanation of the '%p' and '%r' sequences in names
of distributed graph fragments).
.PP
.nf
.fam C
$ mpirun -np 5 dgpart 7 brol%p-%r.dgr brol.map
.fam T
.fi
.SH SEE ALSO
\fBdgtst\fP(1), \fBdgscat\fP(1), \fBamk_grf\fP(1), \fBacpl\fP(1), \fBgmap\fP(1), \fBgmtst\fP(1).
.PP
PT-Scotch user's manual.
.SH AUTHOR
Francois Pellegrini <francois.pellegrini@labri.fr>
|
