HPL_spreadT HPL 2.2 Library Functions February 24, 2016
HPL_spreadT Spread row panel U and forward current column panel.
const enum HPL_SIDE
const int *
const int *
const int *
spreads the local array containing local pieces of U, so
that on exit to this function, a piece of U is contained in every
process row. The array IPLEN contains the number of columns of U,
that should be spread on any given process row. This function also
probes for the presence of the column panel PBCST. If available,
this panel will be forwarded. If PBCST is NULL on input, this
probing mechanism will be disabled.
PBCST (local input/output) HPL_T_panel *
On entry, PBCST points to the data structure containing the
panel (to be broadcast) information.
IFLAG (local input/output) int *
On entry, IFLAG indicates whether or not the broadcast has
already been completed. If not, probing will occur, and the
outcome will be contained in IFLAG on exit.
PANEL (local input/output) HPL_T_panel *
On entry, PANEL points to the data structure containing the
panel (to be spread) information.
SIDE (global input) const enum HPL_SIDE
On entry, SIDE specifies whether the local piece of U located
in process IPMAP[SRCDIST] should be spread to the right or to
the left. This feature is used by the equilibration process.
N (global input) const int
On entry, N specifies the local number of rows of U. N must
be at least zero.
U (local input/output) double *
On entry, U is an array of dimension (LDU,*) containing the
local pieces of U.
LDU (local input) const int
On entry, LDU specifies the local leading dimension of U. LDU
should be at least MAX(1,N).
SRCDIST (local input) const int
On entry, SRCDIST specifies the source process that spreads
its piece of U.
IPLEN (global input) const int *
On entry, IPLEN is an array of dimension NPROW+1. This array
is such that IPLEN[i+1] - IPLEN[i] is the number of rows of U
in each process before process IPMAP[i], with the convention
that IPLEN[nprow] is the total number of rows. In other words
IPLEN[i+1] - IPLEN[i] is the local number of rows of U that
should be moved to process IPMAP[i].
IPMAP (global input) const int *
On entry, IPMAP is an array of dimension NPROW. This array
contains the logarithmic mapping of the processes. In other
words, IPMAP[myrow] is the absolute coordinate of the sorted
IPMAPM1 (global input) const int *
On entry, IPMAPM1 is an array of dimension NPROW. This array
contains the inverse of the logarithmic mapping contained in
IPMAP: For i in [0.. NPROW) IPMAPM1[IPMAP[i]] = i.