1
1
openmpi/ompi/mca/coll/basic/coll_basic_gatherv.c
Rainer Keller 6c5532072a - Split the datatype engine into two parts: an MPI specific part in
OMPI
   and a language agnostic part in OPAL. The convertor is completely
   moved into OPAL.  This offers several benefits as described in RFC
   http://www.open-mpi.org/community/lists/devel/2009/07/6387.php
   namely:
    - Fewer basic types (int* and float* types, boolean and wchar
    - Fixing naming scheme to ompi-nomenclature.
    - Usability outside of the ompi-layer.
 - Due to the fixed nature of simple opal types, their information is
   completely
   known at compile time and therefore constified
 - With fewer datatypes (22), the actual sizes of bit-field types may be
   reduced
   from 64 to 32 bits, allowing reorganizing the opal_datatype
   structure, eliminating holes and keeping data required in convertor
   (upon send/recv) in one cacheline...
   This has implications to the convertor-datastructure and other parts
   of the code.
 - Several performance tests have been run, the netpipe latency does not
   change with
   this patch on Linux/x86-64 on the smoky cluster.
 - Extensive tests have been done to verify correctness (no new
   regressions) using:
   1. mpi_test_suite on linux/x86-64 using clean ompi-trunk and
    ompi-ddt:
    a. running both trunk and ompi-ddt resulted in no differences
       (except for MPI_SHORT_INT and MPI_TYPE_MIX_LB_UB do now run
       correctly).
    b. with --enable-memchecker and running under valgrind (one buglet
       when run with static found in test-suite, commited)
   2. ibm testsuite on linux/x86-64 using clean ompi-trunk and ompi-ddt:
      all passed (except for the dynamic/ tests failed!! as trunk/MTT)
   3. compilation and usage of HDF5 tests on Jaguar using PGI and
      PathScale compilers.
   4. compilation and usage on Scicortex.
 - Please note, that for the heterogeneous case, (-m32 compiled
   binaries/ompi), neither
   ompi-trunk, nor ompi-ddt branch would successfully launch.

This commit was SVN r21641.
2009-07-13 04:56:31 +00:00

160 строки
5.1 KiB
C

/*
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2005 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
* University of Stuttgart. All rights reserved.
* Copyright (c) 2004-2005 The Regents of the University of California.
* All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "ompi_config.h"
#include "coll_basic.h"
#include "mpi.h"
#include "ompi/constants.h"
#include "ompi/datatype/ompi_datatype.h"
#include "ompi/mca/coll/coll.h"
#include "ompi/mca/coll/base/coll_tags.h"
#include "ompi/mca/pml/pml.h"
#include "coll_basic.h"
/*
* gatherv_intra
*
* Function: - basic gatherv operation
* Accepts: - same arguments as MPI_Gatherv()
* Returns: - MPI_SUCCESS or error code
*/
int
mca_coll_basic_gatherv_intra(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void *rbuf, int *rcounts, int *disps,
struct ompi_datatype_t *rdtype, int root,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
int i, rank, size, err;
char *ptmp;
ptrdiff_t lb, extent;
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
/* Everyone but root sends data and returns. Don't send anything
for sendcounts of 0 (even though MPI_Gatherv has a guard for 0
counts, this routine is used elsewhere, like the implementation
of allgatherv, so it's possible to get here with a scount of
0) */
if (rank != root) {
if (scount > 0) {
return MCA_PML_CALL(send(sbuf, scount, sdtype, root,
MCA_COLL_BASE_TAG_GATHERV,
MCA_PML_BASE_SEND_STANDARD, comm));
}
return MPI_SUCCESS;
}
/* I am the root, loop receiving data. */
err = ompi_datatype_get_extent(rdtype, &lb, &extent);
if (OMPI_SUCCESS != err) {
return OMPI_ERROR;
}
for (i = 0; i < size; ++i) {
ptmp = ((char *) rbuf) + (extent * disps[i]);
if (i == rank) {
/* simple optimization */
if (MPI_IN_PLACE != sbuf && (0 < scount) && (0 < rcounts[i])) {
err = ompi_datatype_sndrcv(sbuf, scount, sdtype,
ptmp, rcounts[i], rdtype);
}
} else {
/* Only receive if there is something to receive */
if (rcounts[i] > 0) {
err = MCA_PML_CALL(recv(ptmp, rcounts[i], rdtype, i,
MCA_COLL_BASE_TAG_GATHERV,
comm, MPI_STATUS_IGNORE));
}
}
if (MPI_SUCCESS != err) {
return err;
}
}
/* All done */
return MPI_SUCCESS;
}
/*
* gatherv_inter
*
* Function: - basic gatherv operation
* Accepts: - same arguments as MPI_Gatherv()
* Returns: - MPI_SUCCESS or error code
*/
int
mca_coll_basic_gatherv_inter(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void *rbuf, int *rcounts, int *disps,
struct ompi_datatype_t *rdtype, int root,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
int i, size, err;
char *ptmp;
ptrdiff_t lb, extent;
mca_coll_basic_module_t *basic_module = (mca_coll_basic_module_t*) module;
ompi_request_t **reqs = basic_module->mccb_reqs;
size = ompi_comm_remote_size(comm);
/* If not root, receive data. Note that we will only get here if
* scount > 0 or rank == root. */
if (MPI_PROC_NULL == root) {
/* do nothing */
err = OMPI_SUCCESS;
} else if (MPI_ROOT != root) {
/* Everyone but root sends data and returns. */
err = MCA_PML_CALL(send(sbuf, scount, sdtype, root,
MCA_COLL_BASE_TAG_GATHERV,
MCA_PML_BASE_SEND_STANDARD, comm));
} else {
/* I am the root, loop receiving data. */
err = ompi_datatype_get_extent(rdtype, &lb, &extent);
if (OMPI_SUCCESS != err) {
return OMPI_ERROR;
}
for (i = 0; i < size; ++i) {
ptmp = ((char *) rbuf) + (extent * disps[i]);
err = MCA_PML_CALL(irecv(ptmp, rcounts[i], rdtype, i,
MCA_COLL_BASE_TAG_GATHERV,
comm, &reqs[i]));
if (OMPI_SUCCESS != err) {
return err;
}
}
err = ompi_request_wait_all(size, reqs, MPI_STATUSES_IGNORE);
}
/* All done */
return err;
}