1
1
openmpi/ompi/mca/pml/cm/pml_cm_send.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

198 строки
7.0 KiB
C

/*
* Copyright (c) 2004-2006 The Regents of the University of California.
* All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "ompi_config.h"
#include "opal/prefetch.h"
#include "ompi/datatype/ompi_datatype.h"
#include "ompi/communicator/communicator.h"
#include "pml_cm.h"
#include "pml_cm_sendreq.h"
int
mca_pml_cm_isend_init(void* buf,
size_t count,
ompi_datatype_t* datatype,
int dst,
int tag,
mca_pml_base_send_mode_t sendmode,
ompi_communicator_t* comm,
ompi_request_t** request)
{
int ret;
mca_pml_cm_hvy_send_request_t *sendreq;
ompi_proc_t* ompi_proc;
MCA_PML_CM_HVY_SEND_REQUEST_ALLOC(sendreq, comm, dst, ompi_proc, ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) return ret;
MCA_PML_CM_HVY_SEND_REQUEST_INIT(sendreq, ompi_proc, comm, tag, dst,
datatype, sendmode, true, false, buf, count);
*request = (ompi_request_t*) sendreq;
return OMPI_SUCCESS;
}
int
mca_pml_cm_isend(void* buf,
size_t count,
ompi_datatype_t* datatype,
int dst,
int tag,
mca_pml_base_send_mode_t sendmode,
ompi_communicator_t* comm,
ompi_request_t** request)
{
int ret;
if(sendmode == MCA_PML_BASE_SEND_BUFFERED ) {
mca_pml_cm_hvy_send_request_t* sendreq;
ompi_proc_t* ompi_proc;
MCA_PML_CM_HVY_SEND_REQUEST_ALLOC(sendreq, comm, dst, ompi_proc, ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) return ret;
MCA_PML_CM_HVY_SEND_REQUEST_INIT(sendreq,
ompi_proc,
comm,
tag,
dst,
datatype,
sendmode,
false,
false,
buf,
count);
MCA_PML_CM_HVY_SEND_REQUEST_START( sendreq, ret);
if (OPAL_LIKELY(OMPI_SUCCESS == ret)) *request = (ompi_request_t*) sendreq;
} else {
mca_pml_cm_thin_send_request_t* sendreq;
ompi_proc_t* ompi_proc;
MCA_PML_CM_THIN_SEND_REQUEST_ALLOC(sendreq, comm, dst, ompi_proc, ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) return ret;
MCA_PML_CM_THIN_SEND_REQUEST_INIT(sendreq,
ompi_proc,
comm,
tag,
dst,
datatype,
sendmode,
buf,
count);
MCA_PML_CM_THIN_SEND_REQUEST_START(
sendreq,
comm,
tag,
dst,
sendmode,
false,
ret);
if (OPAL_LIKELY(OMPI_SUCCESS == ret)) *request = (ompi_request_t*) sendreq;
}
return ret;
}
int
mca_pml_cm_send(void *buf,
size_t count,
ompi_datatype_t* datatype,
int dst,
int tag,
mca_pml_base_send_mode_t sendmode,
ompi_communicator_t* comm)
{
int ret = OMPI_ERROR;
if(sendmode == MCA_PML_BASE_SEND_BUFFERED) {
mca_pml_cm_hvy_send_request_t *sendreq;
ompi_proc_t * ompi_proc;
MCA_PML_CM_HVY_SEND_REQUEST_ALLOC(sendreq, comm, dst, ompi_proc, ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) return ret;
MCA_PML_CM_HVY_SEND_REQUEST_INIT(sendreq,
ompi_proc,
comm,
tag,
dst,
datatype,
sendmode,
false,
false,
buf,
count);
MCA_PML_CM_HVY_SEND_REQUEST_START(sendreq, ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) {
MCA_PML_CM_HVY_SEND_REQUEST_RETURN(sendreq);
return ret;
}
ompi_request_free( (ompi_request_t**)&sendreq );
} else {
mca_pml_cm_thin_send_request_t *sendreq;
ompi_proc_t * ompi_proc;
MCA_PML_CM_THIN_SEND_REQUEST_ALLOC(sendreq, comm, dst, ompi_proc, ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) return ret;
MCA_PML_CM_THIN_SEND_REQUEST_INIT(sendreq,
ompi_proc,
comm,
tag,
dst,
datatype,
sendmode,
buf,
count);
if (NULL == ompi_mtl->mtl_send) {
MCA_PML_CM_THIN_SEND_REQUEST_START(sendreq,
comm,
tag,
dst,
sendmode,
false,
ret);
if (OPAL_UNLIKELY(OMPI_SUCCESS != ret)) {
MCA_PML_CM_THIN_SEND_REQUEST_RETURN(sendreq);
return ret;
}
ompi_request_wait_completion(&sendreq->req_send.req_base.req_ompi);
ompi_request_free( (ompi_request_t**)&sendreq );
} else {
MCA_PML_CM_SEND_REQUEST_START_SETUP((&sendreq->req_send));
ret = OMPI_MTL_CALL(send(ompi_mtl,
comm,
dst,
tag,
&sendreq->req_send.req_base.req_convertor,
sendmode));
/* Allow a quick path for the request return */
sendreq->req_send.req_base.req_free_called = true;
MCA_PML_CM_THIN_SEND_REQUEST_PML_COMPLETE(sendreq);
}
}
return ret;
}