1dgscat(1) PT-Scotch user's manual dgscat(1)
2
6 dgscat - build distributed source graph file fragments from a central‐
7 ized source graph file
8
10 dgscat [options] [igfile] [ogfile]
11
13 The dgscat program reads a centralized source graph igfile and writes
14 it back on the form of a set of files ogfile representing fragments of
15 a distributed source graph.
16 When file names are not specified, data is read from standard input and
18 written to standard output. Standard streams can also be explicitly
19 represented by a dash '-'.
20 When the proper libraries have been included at compile time, dgord can
22 directly handle compressed graphs, both as input and output. A stream
23 is treated as compressed whenever its name is postfixed with a com‐
24 pressed file extension, such as in 'brol.grf.bz2' or '-.gz'. The com‐
25 pression formats which can be supported are the bzip2 format ('.bz2'),
26 the gzip format ('.gz'), and the lzma format ('.lzma', on input only).
27 dgord bases on implementations of the MPI interface to spread work
29 across the processing elements. It is therefore not likely to be run
30 directly, but instead through some launcher command such as mpirun.
31
33 In order to tell whether programs should read from, or write to, a sin‐
34 gle file located on only one processor, or to multiple instances of the
35 same file on all of the processors, or else to distinct files on each
36 of the processors, a special grammar has been designed, which is based
37 on the '%' escape character. Four such escape sequences are defined,
38 which are interpreted independently on every processor, prior to file
39 opening. By default, when a filename is provided, it is assumed that
40 the file is to be opened on only one of the processors, called the root
41 processor, which is usually process $0$ of the communicator within
42 which the program is run. The index of the root processor can be
43 changed by means of the -r option. Using any of the first three escape
44 sequences below will instruct programs to open in parallel a file of
45 name equal to the interpreted filename, on every processor on which
46 they are run.
47 %p Replaced by the number of processes in the global communicator
49 in which the program is run. Leads to parallel opening.
50 %r Replaced on each process running the program by the rank of this
52 process in the global communicator. Leads to parallel opening.
53 %- Discarded, but leads to parallel opening. This sequence is
55 mainly used to instruct programs to open on every processor a
56 file of identical name. The opened files can be, according
57 whether the given path leads to a shared directory or to direc‐
58 tories that are local to each processor, either to the opening
59 of multiple instances of the same file, or to the opening of
60 distinct files which may each have a different content, respec‐
61 tively (but in this latter case it is much recommended to iden‐
62 tify files by means of the '%r' sequence).
63 %% Replaced by a single '%' character. File names using this escape
65 sequence are not considered for parallel opening, unless one or
66 several of the three other escape sequences are also present.
67For instance, filename 'brol' will lead to the opening of file 'brol' on the
69root processor only, filename '%-brol' (or even 'br%-ol') will lead to the
70parallel opening of files called 'brol' on every processor, and filename
71'brol%p-%r' will lead to the opening of files on which the program were to
72run.
73
75 -c Check the consistency of the input source graph after loading it
76 into memory.
77 -h Display some help.
79 -rpnum Set root process for centralized files (default is 0).
81 -V Display program version and copyright.
83
85 Run dgscat on 5 processing elements to scatter centralized graph file
86 brol.grf into 5 gzipped file fragments brol5-0.dgr.gz to
87 brol5-4.dgr.gz.
88 $ mpirun -np 5 dgscat brol.grf brol%p-%r.dgr.gz
90
93 dgtst(1), dgord(1), gmk_hy(1).
94 PT-Scotch user's manual.
96
98 Francois Pellegrini <francois.pellegrini@labri.fr>
99 September 08, 2008 dgscat(1)