1
1
openmpi/ompi/mca/coll/base/coll_base_alltoall.c

661 строка
25 KiB
C

/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil -*- */
/*
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2015 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 (c) 2013 Los Alamos National Security, LLC. All Rights
* reserved.
* Copyright (c) 2014-2015 Research Organization for Information Science
* and Technology (RIST). All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "ompi_config.h"
#include "mpi.h"
#include "ompi/constants.h"
#include "ompi/datatype/ompi_datatype.h"
#include "ompi/communicator/communicator.h"
#include "ompi/mca/coll/coll.h"
#include "ompi/mca/coll/base/coll_tags.h"
#include "ompi/mca/pml/pml.h"
#include "ompi/mca/coll/base/coll_base_functions.h"
#include "coll_base_topo.h"
#include "coll_base_util.h"
/* MPI_IN_PLACE all to all algorithm. TODO: implement a better one. */
int
mca_coll_base_alltoall_intra_basic_inplace(void *rbuf, int rcount,
struct ompi_datatype_t *rdtype,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
mca_coll_base_module_t *base_module = (mca_coll_base_module_t*) module;
int i, j, size, rank, err=MPI_SUCCESS;
MPI_Request *preq;
char *tmp_buffer;
size_t max_size;
ptrdiff_t ext;
/* Initialize. */
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
/* If only one process, we're done. */
if (1 == size) {
return MPI_SUCCESS;
}
/* Find the largest receive amount */
ompi_datatype_type_extent (rdtype, &ext);
max_size = ext * rcount;
/* Allocate a temporary buffer */
tmp_buffer = calloc (max_size, 1);
if (NULL == tmp_buffer) {
return OMPI_ERR_OUT_OF_RESOURCE;
}
/* in-place alltoall slow algorithm (but works) */
for (i = 0 ; i < size ; ++i) {
for (j = i+1 ; j < size ; ++j) {
/* Initiate all send/recv to/from others. */
preq = coll_base_comm_get_reqs(base_module->base_data, size * 2);
if (i == rank) {
/* Copy the data into the temporary buffer */
err = ompi_datatype_copy_content_same_ddt (rdtype, rcount, tmp_buffer,
(char *) rbuf + j * max_size);
if (MPI_SUCCESS != err) { goto error_hndl; }
/* Exchange data with the peer */
err = MCA_PML_CALL(irecv ((char *) rbuf + max_size * j, rcount, rdtype,
j, MCA_COLL_BASE_TAG_ALLTOALL, comm, preq++));
if (MPI_SUCCESS != err) { goto error_hndl; }
err = MCA_PML_CALL(isend ((char *) tmp_buffer, rcount, rdtype,
j, MCA_COLL_BASE_TAG_ALLTOALL, MCA_PML_BASE_SEND_STANDARD,
comm, preq++));
if (MPI_SUCCESS != err) { goto error_hndl; }
} else if (j == rank) {
/* Copy the data into the temporary buffer */
err = ompi_datatype_copy_content_same_ddt (rdtype, rcount, tmp_buffer,
(char *) rbuf + i * max_size);
if (MPI_SUCCESS != err) { goto error_hndl; }
/* Exchange data with the peer */
err = MCA_PML_CALL(irecv ((char *) rbuf + max_size * i, rcount, rdtype,
i, MCA_COLL_BASE_TAG_ALLTOALL, comm, preq++));
if (MPI_SUCCESS != err) { goto error_hndl; }
err = MCA_PML_CALL(isend ((char *) tmp_buffer, rcount, rdtype,
i, MCA_COLL_BASE_TAG_ALLTOALL, MCA_PML_BASE_SEND_STANDARD,
comm, preq++));
if (MPI_SUCCESS != err) { goto error_hndl; }
} else {
continue;
}
/* Wait for the requests to complete */
err = ompi_request_wait_all (2, base_module->base_data->mcct_reqs, MPI_STATUSES_IGNORE);
if (MPI_SUCCESS != err) { goto error_hndl; }
}
}
error_hndl:
/* Free the temporary buffer */
free (tmp_buffer);
/* All done */
return err;
}
int ompi_coll_base_alltoall_intra_pairwise(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void* rbuf, int rcount,
struct ompi_datatype_t *rdtype,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
int line = -1, err = 0, rank, size, step, sendto, recvfrom;
void * tmpsend, *tmprecv;
ptrdiff_t lb, sext, rext;
if (MPI_IN_PLACE == sbuf) {
return mca_coll_base_alltoall_intra_basic_inplace (rbuf, rcount, rdtype,
comm, module);
}
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"coll:base:alltoall_intra_pairwise rank %d", rank));
err = ompi_datatype_get_extent (sdtype, &lb, &sext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
err = ompi_datatype_get_extent (rdtype, &lb, &rext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* Perform pairwise exchange - starting from 1 so the local copy is last */
for (step = 1; step < size + 1; step++) {
/* Determine sender and receiver for this step. */
sendto = (rank + step) % size;
recvfrom = (rank + size - step) % size;
/* Determine sending and receiving locations */
tmpsend = (char*)sbuf + (ptrdiff_t)sendto * sext * (ptrdiff_t)scount;
tmprecv = (char*)rbuf + (ptrdiff_t)recvfrom * rext * (ptrdiff_t)rcount;
/* send and receive */
err = ompi_coll_base_sendrecv( tmpsend, scount, sdtype, sendto,
MCA_COLL_BASE_TAG_ALLTOALL,
tmprecv, rcount, rdtype, recvfrom,
MCA_COLL_BASE_TAG_ALLTOALL,
comm, MPI_STATUS_IGNORE, rank);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
}
return MPI_SUCCESS;
err_hndl:
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"%s:%4d\tError occurred %d, rank %2d", __FILE__, line,
err, rank));
return err;
}
int ompi_coll_base_alltoall_intra_bruck(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void* rbuf, int rcount,
struct ompi_datatype_t *rdtype,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
int i, k, line = -1, rank, size, err = 0;
int sendto, recvfrom, distance, *displs = NULL, *blen = NULL;
char *tmpbuf = NULL, *tmpbuf_free = NULL;
ptrdiff_t rlb, slb, tlb, sext, rext, tsext;
struct ompi_datatype_t *new_ddt;
if (MPI_IN_PLACE == sbuf) {
return mca_coll_base_alltoall_intra_basic_inplace (rbuf, rcount, rdtype,
comm, module);
}
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"coll:base:alltoall_intra_bruck rank %d", rank));
err = ompi_datatype_get_extent (sdtype, &slb, &sext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
err = ompi_datatype_get_true_extent(sdtype, &tlb, &tsext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
err = ompi_datatype_get_extent (rdtype, &rlb, &rext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
displs = (int *) malloc(size * sizeof(int));
if (displs == NULL) { line = __LINE__; err = -1; goto err_hndl; }
blen = (int *) malloc(size * sizeof(int));
if (blen == NULL) { line = __LINE__; err = -1; goto err_hndl; }
/* tmp buffer allocation for message data */
tmpbuf_free = (char *) malloc(tsext + ((ptrdiff_t)scount * (ptrdiff_t)size - 1) * sext);
if (tmpbuf_free == NULL) { line = __LINE__; err = -1; goto err_hndl; }
tmpbuf = tmpbuf_free - slb;
/* Step 1 - local rotation - shift up by rank */
err = ompi_datatype_copy_content_same_ddt (sdtype,
(int32_t) ((ptrdiff_t)(size - rank) * (ptrdiff_t)scount),
tmpbuf,
((char*) sbuf) + (ptrdiff_t)rank * (ptrdiff_t)scount * sext);
if (err<0) {
line = __LINE__; err = -1; goto err_hndl;
}
if (rank != 0) {
err = ompi_datatype_copy_content_same_ddt (sdtype, (ptrdiff_t)rank * (ptrdiff_t)scount,
tmpbuf + (ptrdiff_t)(size - rank) * (ptrdiff_t)scount* sext,
(char*) sbuf);
if (err<0) {
line = __LINE__; err = -1; goto err_hndl;
}
}
/* perform communication step */
for (distance = 1; distance < size; distance<<=1) {
sendto = (rank + distance) % size;
recvfrom = (rank - distance + size) % size;
k = 0;
/* create indexed datatype */
for (i = 1; i < size; i++) {
if (( i & distance) == distance) {
displs[k] = (ptrdiff_t)i * (ptrdiff_t)scount;
blen[k] = scount;
k++;
}
}
/* Set indexes and displacements */
err = ompi_datatype_create_indexed(k, blen, displs, sdtype, &new_ddt);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* Commit the new datatype */
err = ompi_datatype_commit(&new_ddt);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* Sendreceive */
err = ompi_coll_base_sendrecv ( tmpbuf, 1, new_ddt, sendto,
MCA_COLL_BASE_TAG_ALLTOALL,
rbuf, 1, new_ddt, recvfrom,
MCA_COLL_BASE_TAG_ALLTOALL,
comm, MPI_STATUS_IGNORE, rank );
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* Copy back new data from recvbuf to tmpbuf */
err = ompi_datatype_copy_content_same_ddt(new_ddt, 1,tmpbuf, (char *) rbuf);
if (err < 0) { line = __LINE__; err = -1; goto err_hndl; }
/* free ddt */
err = ompi_datatype_destroy(&new_ddt);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
} /* end of for (distance = 1... */
/* Step 3 - local rotation - */
for (i = 0; i < size; i++) {
err = ompi_datatype_copy_content_same_ddt (rdtype, (int32_t) rcount,
((char*)rbuf) + ((ptrdiff_t)((rank - i + size) % size) * (ptrdiff_t)rcount * rext),
tmpbuf + (ptrdiff_t)i * (ptrdiff_t)rcount * rext);
if (err < 0) { line = __LINE__; err = -1; goto err_hndl; }
}
/* Step 4 - clean up */
if (tmpbuf != NULL) free(tmpbuf_free);
if (displs != NULL) free(displs);
if (blen != NULL) free(blen);
return OMPI_SUCCESS;
err_hndl:
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"%s:%4d\tError occurred %d, rank %2d", __FILE__, line, err,
rank));
if (tmpbuf != NULL) free(tmpbuf_free);
if (displs != NULL) free(displs);
if (blen != NULL) free(blen);
return err;
}
/*
* alltoall_intra_linear_sync
*
* Function: Linear implementation of alltoall with limited number
* of outstanding requests.
* Accepts: Same as MPI_Alltoall(), and the maximum number of
* outstanding requests (actual number is 2 * max, since
* we count receive and send requests separately).
* Returns: MPI_SUCCESS or error code
*
* Description: Algorithm is the following:
* 1) post K irecvs, K <= N
* 2) post K isends, K <= N
* 3) while not done
* - wait for any request to complete
* - replace that request by the new one of the same type.
*/
int ompi_coll_base_alltoall_intra_linear_sync(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void* rbuf, int rcount,
struct ompi_datatype_t *rdtype,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module,
int max_outstanding_reqs)
{
int line, error, ri, si, rank, size, nreqs, nrreqs, nsreqs, total_reqs;
char *psnd, *prcv;
ptrdiff_t slb, sext, rlb, rext;
ompi_request_t **reqs = NULL;
if (MPI_IN_PLACE == sbuf) {
return mca_coll_base_alltoall_intra_basic_inplace (rbuf, rcount, rdtype,
comm, module);
}
/* Initialize. */
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"ompi_coll_base_alltoall_intra_linear_sync rank %d", rank));
error = ompi_datatype_get_extent(sdtype, &slb, &sext);
if (OMPI_SUCCESS != error) {
return error;
}
sext *= scount;
error = ompi_datatype_get_extent(rdtype, &rlb, &rext);
if (OMPI_SUCCESS != error) {
return error;
}
rext *= rcount;
/* simple optimization */
psnd = ((char *) sbuf) + (ptrdiff_t)rank * sext;
prcv = ((char *) rbuf) + (ptrdiff_t)rank * rext;
error = ompi_datatype_sndrcv(psnd, scount, sdtype, prcv, rcount, rdtype);
if (MPI_SUCCESS != error) {
return error;
}
/* If only one process, we're done. */
if (1 == size) {
return MPI_SUCCESS;
}
/* Initiate send/recv to/from others. */
total_reqs = (((max_outstanding_reqs > (size - 1)) ||
(max_outstanding_reqs <= 0)) ?
(size - 1) : (max_outstanding_reqs));
reqs = (ompi_request_t**) malloc( 2 * total_reqs *
sizeof(ompi_request_t*));
if (NULL == reqs) { error = -1; line = __LINE__; goto error_hndl; }
prcv = (char *) rbuf;
psnd = (char *) sbuf;
/* Post first batch or ireceive and isend requests */
for (nreqs = 0, nrreqs = 0, ri = (rank + 1) % size; nreqs < total_reqs;
ri = (ri + 1) % size, ++nreqs, ++nrreqs) {
error =
MCA_PML_CALL(irecv
(prcv + (ptrdiff_t)ri * rext, rcount, rdtype, ri,
MCA_COLL_BASE_TAG_ALLTOALL, comm, &reqs[nreqs]));
if (MPI_SUCCESS != error) { line = __LINE__; goto error_hndl; }
}
for ( nsreqs = 0, si = (rank + size - 1) % size; nreqs < 2 * total_reqs;
si = (si + size - 1) % size, ++nreqs, ++nsreqs) {
error =
MCA_PML_CALL(isend
(psnd + (ptrdiff_t)si * sext, scount, sdtype, si,
MCA_COLL_BASE_TAG_ALLTOALL,
MCA_PML_BASE_SEND_STANDARD, comm, &reqs[nreqs]));
if (MPI_SUCCESS != error) { line = __LINE__; goto error_hndl; }
}
/* Wait for requests to complete */
if (nreqs == 2 * (size - 1)) {
/* Optimization for the case when all requests have been posted */
error = ompi_request_wait_all(nreqs, reqs, MPI_STATUSES_IGNORE);
if (MPI_SUCCESS != error) { line = __LINE__; goto error_hndl; }
} else {
/* As requests complete, replace them with corresponding requests:
- wait for any request to complete, mark the request as
MPI_REQUEST_NULL
- If it was a receive request, replace it with new irecv request
(if any)
- if it was a send request, replace it with new isend request (if any)
*/
int ncreqs = 0;
while (ncreqs < 2 * (size - 1)) {
int completed;
error = ompi_request_wait_any(2 * total_reqs, reqs, &completed,
MPI_STATUS_IGNORE);
if (MPI_SUCCESS != error) { line = __LINE__; goto error_hndl; }
reqs[completed] = MPI_REQUEST_NULL;
ncreqs++;
if (completed < total_reqs) {
if (nrreqs < (size - 1)) {
error =
MCA_PML_CALL(irecv
(prcv + (ptrdiff_t)ri * rext, rcount, rdtype, ri,
MCA_COLL_BASE_TAG_ALLTOALL, comm,
&reqs[completed]));
if (MPI_SUCCESS != error) { line = __LINE__; goto error_hndl; }
++nrreqs;
ri = (ri + 1) % size;
}
} else {
if (nsreqs < (size - 1)) {
error = MCA_PML_CALL(isend
(psnd + (ptrdiff_t)si * sext, scount, sdtype, si,
MCA_COLL_BASE_TAG_ALLTOALL,
MCA_PML_BASE_SEND_STANDARD, comm,
&reqs[completed]));
++nsreqs;
si = (si + size - 1) % size;
}
}
}
}
/* Free the reqs */
free(reqs);
/* All done */
return MPI_SUCCESS;
error_hndl:
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"%s:%4d\tError occurred %d, rank %2d", __FILE__, line, error,
rank));
if (NULL != reqs) free(reqs);
return error;
}
int ompi_coll_base_alltoall_intra_two_procs(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void* rbuf, int rcount,
struct ompi_datatype_t *rdtype,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
int line = -1, err = 0, rank, remote;
void * tmpsend, *tmprecv;
ptrdiff_t sext, rext, lb;
if (MPI_IN_PLACE == sbuf) {
return mca_coll_base_alltoall_intra_basic_inplace (rbuf, rcount, rdtype,
comm, module);
}
rank = ompi_comm_rank(comm);
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"ompi_coll_base_alltoall_intra_two_procs rank %d", rank));
err = ompi_datatype_get_extent (sdtype, &lb, &sext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
err = ompi_datatype_get_extent (rdtype, &lb, &rext);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* exchange data */
remote = rank ^ 1;
tmpsend = (char*)sbuf + (ptrdiff_t)remote * sext * (ptrdiff_t)scount;
tmprecv = (char*)rbuf + (ptrdiff_t)remote * rext * (ptrdiff_t)rcount;
/* send and receive */
err = ompi_coll_base_sendrecv ( tmpsend, scount, sdtype, remote,
MCA_COLL_BASE_TAG_ALLTOALL,
tmprecv, rcount, rdtype, remote,
MCA_COLL_BASE_TAG_ALLTOALL,
comm, MPI_STATUS_IGNORE, rank );
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* ddt sendrecv your own data */
err = ompi_datatype_sndrcv((char*) sbuf + (ptrdiff_t)rank * sext * (ptrdiff_t)scount,
(int32_t) scount, sdtype,
(char*) rbuf + (ptrdiff_t)rank * rext * (ptrdiff_t)rcount,
(int32_t) rcount, rdtype);
if (err != MPI_SUCCESS) { line = __LINE__; goto err_hndl; }
/* done */
return MPI_SUCCESS;
err_hndl:
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"%s:%4d\tError occurred %d, rank %2d", __FILE__, line, err,
rank));
return err;
}
/*
* Linear functions are copied from the BASIC coll module
* they do not segment the message and are simple implementations
* but for some small number of nodes and/or small data sizes they
* are just as fast as base/tree based segmenting operations
* and as such may be selected by the decision functions
* These are copied into this module due to the way we select modules
* in V1. i.e. in V2 we will handle this differently and so will not
* have to duplicate code.
* GEF Oct05 after asking Jeff.
*/
/* copied function (with appropriate renaming) starts here */
int ompi_coll_base_alltoall_intra_basic_linear(void *sbuf, int scount,
struct ompi_datatype_t *sdtype,
void* rbuf, int rcount,
struct ompi_datatype_t *rdtype,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
int i, rank, size, err, nreqs;
char *psnd, *prcv;
MPI_Aint lb, sndinc, rcvinc;
ompi_request_t **req, **sreq, **rreq;
mca_coll_base_module_t *base_module = (mca_coll_base_module_t*) module;
mca_coll_base_comm_t *data = base_module->base_data;
if (MPI_IN_PLACE == sbuf) {
return mca_coll_base_alltoall_intra_basic_inplace (rbuf, rcount, rdtype,
comm, module);
}
/* Initialize. */
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
OPAL_OUTPUT((ompi_coll_base_framework.framework_output,
"ompi_coll_base_alltoall_intra_basic_linear rank %d", rank));
err = ompi_datatype_get_extent(sdtype, &lb, &sndinc);
if (OMPI_SUCCESS != err) {
return err;
}
sndinc *= scount;
err = ompi_datatype_get_extent(rdtype, &lb, &rcvinc);
if (OMPI_SUCCESS != err) {
return err;
}
rcvinc *= rcount;
/* simple optimization */
psnd = ((char *) sbuf) + (ptrdiff_t)rank * sndinc;
prcv = ((char *) rbuf) + (ptrdiff_t)rank * rcvinc;
err = ompi_datatype_sndrcv(psnd, scount, sdtype, prcv, rcount, rdtype);
if (MPI_SUCCESS != err) {
return err;
}
/* If only one process, we're done. */
if (1 == size) {
return MPI_SUCCESS;
}
/* Initiate all send/recv to/from others. */
req = rreq = coll_base_comm_get_reqs(data, (size - 1) * 2);
prcv = (char *) rbuf;
psnd = (char *) sbuf;
/* Post all receives first -- a simple optimization */
for (nreqs = 0, i = (rank + 1) % size; i != rank;
i = (i + 1) % size, ++rreq, ++nreqs) {
err = MCA_PML_CALL(irecv_init
(prcv + (ptrdiff_t)i * rcvinc, rcount, rdtype, i,
MCA_COLL_BASE_TAG_ALLTOALL, comm, rreq));
if (MPI_SUCCESS != err) {
ompi_coll_base_free_reqs(req, nreqs);
return err;
}
}
/* Now post all sends in reverse order
- We would like to minimize the search time through message queue
when messages actually arrive in the order in which they were posted.
*/
sreq = rreq;
for (i = (rank + size - 1) % size; i != rank;
i = (i + size - 1) % size, ++sreq, ++nreqs) {
err = MCA_PML_CALL(isend_init
(psnd + (ptrdiff_t)i * sndinc, scount, sdtype, i,
MCA_COLL_BASE_TAG_ALLTOALL,
MCA_PML_BASE_SEND_STANDARD, comm, sreq));
if (MPI_SUCCESS != err) {
ompi_coll_base_free_reqs(req, nreqs);
return err;
}
}
/* Start your engines. This will never return an error. */
MCA_PML_CALL(start(nreqs, req));
/* Wait for them all. If there's an error, note that we don't
* care what the error was -- just that there *was* an error. The
* PML will finish all requests, even if one or more of them fail.
* i.e., by the end of this call, all the requests are free-able.
* So free them anyway -- even if there was an error, and return
* the error after we free everything. */
err = ompi_request_wait_all(nreqs, req, MPI_STATUSES_IGNORE);
/* Free the reqs */
ompi_coll_base_free_reqs(req, nreqs);
/* All done */
return err;
}
/* copied function (with appropriate renaming) ends here */