1
1
openmpi/ompi/mca/coll/tuned/coll_tuned_util.h

130 строки
5.4 KiB
C
Исходник Обычный вид История

/*
* 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) 2004-2007 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
*
* $HEADER$
*/
#ifndef MCA_COLL_TUNED_UTIL_EXPORT_H
#define MCA_COLL_TUNED_UTIL_EXPORT_H
#include "ompi_config.h"
#include "mpi.h"
#include "opal/mca/mca.h"
- 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 08:56:31 +04:00
#include "ompi/datatype/ompi_datatype.h"
#include "ompi/request/request.h"
#include "ompi/mca/pml/pml.h"
#if defined(c_plusplus) || defined(__cplusplus)
extern "C" {
#endif
/* prototypes */
int ompi_coll_tuned_sendrecv_actual( void* sendbuf, int scount,
ompi_datatype_t* sdatatype,
int dest, int stag,
void* recvbuf, int rcount,
ompi_datatype_t* rdatatype,
int source, int rtag,
struct ompi_communicator_t* comm,
ompi_status_public_t* status );
/* inline functions */
static inline int
ompi_coll_tuned_sendrecv( void* sendbuf, int scount, ompi_datatype_t* sdatatype,
int dest, int stag,
void* recvbuf, int rcount, ompi_datatype_t* rdatatype,
int source, int rtag,
struct ompi_communicator_t* comm,
ompi_status_public_t* status, int myid )
{
if ((dest == myid) && (source == myid)) {
- 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 08:56:31 +04:00
return (int) ompi_datatype_sndrcv(sendbuf, (int32_t) scount, sdatatype,
recvbuf, (int32_t) rcount, rdatatype);
}
return ompi_coll_tuned_sendrecv_actual (sendbuf, scount, sdatatype,
dest, stag,
recvbuf, rcount, rdatatype,
source, rtag, comm, status);
}
int
ompi_coll_tuned_sendrecv_actual_localcompleted( void* sendbuf, int scount,
ompi_datatype_t* sdatatype,
int dest, int stag,
void* recvbuf, int rcount,
ompi_datatype_t* rdatatype,
int source, int rtag,
struct ompi_communicator_t* comm,
ompi_status_public_t* status );
/* inline functions */
static inline int
ompi_coll_tuned_sendrecv_localcompleted( void* sendbuf, int scount,
ompi_datatype_t* sdatatype,
int dest, int stag,
void* recvbuf, int rcount,
ompi_datatype_t* rdatatype,
int source, int rtag,
struct ompi_communicator_t* comm,
ompi_status_public_t* status, int myid )
{
if ((dest == myid) && (source == myid)) {
- 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 08:56:31 +04:00
return (int) ompi_datatype_sndrcv(sendbuf, (int32_t) scount, sdatatype,
recvbuf, (int32_t) rcount, rdatatype);
}
return ompi_coll_tuned_sendrecv_actual_localcompleted (sendbuf, scount,
sdatatype, dest,
stag,
recvbuf, rcount,
rdatatype,
source, rtag, comm,
status);
}
/* inline functions */
static inline int
ompi_coll_tuned_isendrecv( void* sendbuf, int scount, ompi_datatype_t* sdtype,
int dest, int stag, ompi_request_t** sreq,
void* recvbuf, int rcount, ompi_datatype_t* rdtype,
int source, int rtag, ompi_request_t** rreq,
struct ompi_communicator_t* comm ) {
int ret, line;
ret = MCA_PML_CALL(irecv(recvbuf, rcount, rdtype, source, rtag, comm, rreq));
if (MPI_SUCCESS != ret) { line = __LINE__; goto error_handler; }
ret = MCA_PML_CALL(isend(sendbuf, scount, sdtype, dest, stag,
MCA_PML_BASE_SEND_STANDARD, comm, sreq));
if (MPI_SUCCESS != ret) { line = __LINE__; goto error_handler; }
return MPI_SUCCESS;
error_handler:
OPAL_OUTPUT((ompi_coll_tuned_stream, "%s:%d\tError occurred %d\n",
__FILE__, line, ret));
return ret;
}
#if defined(c_plusplus) || defined(__cplusplus)
}
#endif
#endif /* MCA_COLL_TUNED_UTIL_EXPORT_H */