Statistiques
| Révision :

root / man / man3 / HPL_equil.3 @ 1

Historique | Voir | Annoter | Télécharger (2,82 ko)

1 
.TH HPL_equil 3 "September 10, 2008" "HPL 2.0" "HPL Library Functions"
2 
.SH NAME
3 
HPL_equil \- Equilibrate U and forward the column panel L.
4 
.SH SYNOPSIS
5 
\fB\&#include "hpl.h"\fR
6
7 
\fB\&void\fR
8 
\fB\&HPL_equil(\fR
9 
\fB\&HPL_T_panel *\fR
10 
\fI\&PBCST\fR,
11 
\fB\&int *\fR
12 
\fI\&IFLAG\fR,
13 
\fB\&HPL_T_panel *\fR
14 
\fI\&PANEL\fR,
15 
\fB\&const enum HPL_TRANS\fR
16 
\fI\&TRANS\fR,
17 
\fB\&const int\fR
18 
\fI\&N\fR,
19 
\fB\&double *\fR
20 
\fI\&U\fR,
21 
\fB\&const int\fR
22 
\fI\&LDU\fR,
23 
\fB\&int *\fR
24 
\fI\&IPLEN\fR,
25 
\fB\&const int *\fR
26 
\fI\&IPMAP\fR,
27 
\fB\&const int *\fR
28 
\fI\&IPMAPM1\fR,
29 
\fB\&int *\fR
30 
\fI\&IWORK\fR
31 
\fB\&);\fR
32 
.SH DESCRIPTION
33 
\fB\&HPL_equil\fR
34 
equilibrates  the  local  pieces  of U, so that on exit to
35 
this function, pieces of U contained in every process row are of the
36 
same size. This phase makes the rolling phase optimal.  In addition,
37 
this  function probes  for  the  column panel L and forwards it when
38 
possible.
39 
.SH ARGUMENTS
40 
.TP 8
41 
PBCST   (local input/output)    HPL_T_panel *
42 
On entry,  PBCST  points to the data structure containing the
43 
panel (to be broadcast) information.
44 
.TP 8
45 
IFLAG   (local input/output)    int *
46 
On entry, IFLAG  indicates  whether or not  the broadcast has
47 
already been completed.  If not,  probing will occur, and the
48 
outcome will be contained in IFLAG on exit.
49 
.TP 8
50 
PANEL   (local input/output)    HPL_T_panel *
51 
On entry,  PANEL  points to the data structure containing the
52 
panel (to be equilibrated) information.
53 
.TP 8
54 
TRANS   (global input)          const enum HPL_TRANS
55 
On entry, TRANS specifies whether  U  is stored in transposed
56 
or non-transposed form.
57 
.TP 8
58 
N       (local input)           const int
59 
On entry, N  specifies the number of rows or columns of  U. N
60 
must be at least 0.
61 
.TP 8
62 
U       (local input/output)    double *
63 
On entry,  U  is an array of dimension (LDU,*) containing the
64 
local pieces of U in each process row.
65 
.TP 8
66 
LDU     (local input)           const int
67 
On entry, LDU specifies the local leading dimension of U. LDU
68 
should be at least MAX(1,IPLEN[nprow]) when  U  is stored  in
69 
non-transposed form, and MAX(1,N) otherwise.
70 
.TP 8
71 
IPLEN   (global input)          int *
72 
On entry, IPLEN is an array of dimension NPROW+1.  This array
73 
is such that IPLEN[i+1] - IPLEN[i] is the number of rows of U
74 
in process IPMAP[i].
75 
.TP 8
76 
IPMAP   (global input)          const int *
77 
On entry, IPMAP is an array of dimension  NPROW.  This  array
78 
contains  the  logarithmic mapping of the processes. In other
79 
words, IPMAP[myrow]  is the absolute coordinate of the sorted
80 
process.
81 
.TP 8
82 
IPMAPM1 (global input)          const int *
83 
On entry, IPMAPM1  is an array of dimension NPROW. This array
84 
contains  the inverse of the logarithmic mapping contained in
85 
IPMAP: For i in [0.. NPROCS) IPMAPM1[IPMAP[i]] = i.
86 
.TP 8
87 
IWORK   (workspace)             int *
88 
On entry, IWORK is a workarray of dimension NPROW+1.
89 
.SH SEE ALSO
90 
.BR HPL_pdlaswp01N \ (3),
91 
.BR HPL_pdlaswp01T \ (3).