HPL_plindx1 - Man Page

#include "hpl.h"
 void HPL_plindx1( HPL_T_panel * PANEL, const int K, const int * IPID, int * IPA, int * LINDXA, int * LINDXAU, int * IPLEN, int * IPMAP, int * IPMAPM1, int * PERMU, int * IWORK );

HPL_plindx1 computes two local arrays  LINDXA and  LINDXAU  containing the  local  source and final destination position  resulting from the application of row interchanges.  In addition, this function computes three arrays IPLEN, IPMAP and IPMAPM1  that contain  the  logarithmic mapping information for the spreading phase.

PANEL   (local input/output)    HPL_T_panel *

On entry,  PANEL  points to the data structure containing the panel information.

K       (global input)          const int

On entry, K specifies the number of entries in IPID.  K is at least 2*N, and at most 4*N.

IPID    (global input)          const int *

On entry,  IPID  is an array of length K. The first K entries of that array contain the src and final destination resulting from the application of the interchanges.

IPA     (global output)         int *

On exit,  IPA  specifies  the number of rows that the current process row has that either belong to U  or should be swapped with remote rows of A.

LINDXA  (global output)         int *

On entry, LINDXA  is an array of dimension 2*N. On exit, this array contains the local indexes of the rows of A I have that should be copied into U.

LINDXAU (global output)         int *

On exit, LINDXAU  is an array of dimension 2*N. On exit, this array contains  the local destination  information encoded as follows.  If LINDXAU(k) >= 0, row  LINDXA(k)  of A  is  to be copied in U at position LINDXAU(k).  Otherwise, row LINDXA(k) of A should be locally copied into A(-LINDXAU(k),:).

IPLEN   (global output)         int *

On entry, IPLEN is an array of dimension NPROW + 1. On  exit, this array is such that  IPLEN[i]  is the number of rows of A in  the  processes  before  process  IPMAP[i]  after the sort with the convention that IPLEN[nprow]  is the total number of rows of the panel.  In other words IPLEN[i+1]-IPLEN[i] is the local number of rows of A that should be moved to the process IPMAP[i]. IPLEN is such that the number of rows of the source process  row can be computed as  IPLEN[1] - IPLEN[0], and the remaining  entries  of  this  array  are  sorted  so that the quantities IPLEN[i+1] - IPLEN[i] are logarithmically sorted.

IPMAP   (global output)         int *

On entry, IPMAP is an array of dimension NPROW. On exit, this array contains  the logarithmic mapping of the processes.  In other words, IPMAP[myrow] is the corresponding sorted process coordinate.

IPMAPM1 (global output)         int *

On entry, IPMAPM1  is an array of dimension NPROW.  On  exit, this  array  contains  the inverse of the logarithmic mapping contained  in  IPMAP:  IPMAPM1[ IPMAP[i] ] = i,  for all i in [0.. NPROCS)

PERMU   (global output)         int *

On entry,  PERMU  is an array of dimension JB. On exit, PERMU contains  a sequence of permutations,  that should be applied in increasing order to permute in place the row panel U.

IWORK   (workspace)             int *

On entry, IWORK is a workarray of dimension 2*JB.

HPL_pdlaswp00N (3), HPL_pdlaswp00T (3), HPL_pdlaswp01N (3), HPL_pdlaswp01T (3).

HPL_logsort(3), HPL_pdlaswp01N(3), HPL_pdlaswp01T(3), HPL_perm(3).