16b49dc5b3
Among many other things: - Fix an imbalance bug in MPI_allgather - Accept more human readable configuration files. We can now specify the collective by name instead of a magic number, and the component we want to use also by name. - Add the capability to have optional arguments in the collective communication configuration file. Right now the capability exists for segment lengths, but is yet to be connected with the algorithms. - Redo the initialization of all HAN collectives. Cleanup the fallback collective support. - In case the module is unable to deliver the expected result, it will fallback executing the collective operation on another collective component. This change make the support for this fallback simpler to use. - Implement a fallback allowing a HAN module to remove itself as potential active collective module, and instead fallback to the next module in line. - Completely disable the HAN modules on error. From the moment an error is encountered they remove themselves from the communicator, and in case some other modules calls them simply behave as a pass-through. Communicator: provide ompi_comm_split_with_info to split and provide info at the same time Add ompi_comm_coll_preference info key to control collective component selection COLL HAN: use info keys instead of component-level variable to communicate topology level between abstraction layers - The info value is a comma-separated list of entries, which are chosen with decreasing priorities. This overrides the priority of the component, unless the component has disqualified itself. An entry prefixed with ^ starts the ignore-list. Any entry following this character will be ingnored during the collective component selection for the communicator. Example: "sm,libnbc,^han,adapt" gives sm the highest preference, followed by libnbc. The components han and adapt are ignored in the selection process. - Allocate a temporary buffer for all lower-level leaders (length 2 segments) - Fix the handling of MPI_IN_PLACE for gather and scatter. COLL HAN: Fix topology handling - HAN should not rely on node names to determine the ordering of ranks. Instead, use the node leaders as identifiers and short-cut if the node-leaders agree that ranks are consecutive. Also, error out if the rank distribution is imbalanced for now. Signed-off-by: Xi Luo <xluo12@vols.utk.edu> Signed-off-by: Joseph Schuchart <schuchart@icl.utk.edu> Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
259 строки
11 KiB
C
259 строки
11 KiB
C
/*
|
|
* Copyright (c) 2018-2020 The University of Tennessee and The University
|
|
* of Tennessee Research Foundation. All rights
|
|
* reserved.
|
|
* $COPYRIGHT$
|
|
*
|
|
* Additional copyrights may follow
|
|
*
|
|
* $HEADER$
|
|
*/
|
|
|
|
#include "coll_han.h"
|
|
#include "ompi/mca/coll/base/coll_base_functions.h"
|
|
#include "ompi/mca/coll/base/coll_tags.h"
|
|
#include "ompi/mca/pml/pml.h"
|
|
#include "coll_han_trigger.h"
|
|
|
|
static int mca_coll_han_scatter_us_task(void *task_args);
|
|
static int mca_coll_han_scatter_ls_task(void *task_args);
|
|
|
|
/* Only work with regular situation (each node has equal number of processes) */
|
|
|
|
static inline void
|
|
mca_coll_han_set_scatter_args(mca_coll_han_scatter_args_t * args,
|
|
mca_coll_task_t * cur_task,
|
|
void *sbuf,
|
|
void *sbuf_inter_free,
|
|
void *sbuf_reorder_free,
|
|
int scount,
|
|
struct ompi_datatype_t *sdtype,
|
|
void *rbuf,
|
|
int rcount,
|
|
struct ompi_datatype_t *rdtype,
|
|
int root,
|
|
int root_up_rank,
|
|
int root_low_rank,
|
|
struct ompi_communicator_t *up_comm,
|
|
struct ompi_communicator_t *low_comm,
|
|
int w_rank, bool noop, ompi_request_t * req)
|
|
{
|
|
args->cur_task = cur_task;
|
|
args->sbuf = sbuf;
|
|
args->sbuf_inter_free = sbuf_inter_free;
|
|
args->sbuf_reorder_free = sbuf_reorder_free;
|
|
args->scount = scount;
|
|
args->sdtype = sdtype;
|
|
args->rbuf = rbuf;
|
|
args->rcount = rcount;
|
|
args->rdtype = rdtype;
|
|
args->root = root;
|
|
args->root_up_rank = root_up_rank;
|
|
args->root_low_rank = root_low_rank;
|
|
args->up_comm = up_comm;
|
|
args->low_comm = low_comm;
|
|
args->w_rank = w_rank;
|
|
args->noop = noop;
|
|
args->req = req;
|
|
}
|
|
|
|
int
|
|
mca_coll_han_scatter_intra(const void *sbuf, int scount,
|
|
struct ompi_datatype_t *sdtype,
|
|
void *rbuf, int rcount,
|
|
struct ompi_datatype_t *rdtype,
|
|
int root,
|
|
struct ompi_communicator_t *comm, mca_coll_base_module_t * module)
|
|
{
|
|
mca_coll_han_module_t *han_module = (mca_coll_han_module_t *) module;
|
|
int i, j, w_rank, w_size;
|
|
w_rank = ompi_comm_rank(comm);
|
|
w_size = ompi_comm_size(comm);
|
|
|
|
/* Create the subcommunicators */
|
|
if( OMPI_SUCCESS != mca_coll_han_comm_create(comm, han_module) ) { /* Let's hope the error is consistently returned across the entire communicator */
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output,
|
|
"han cannot handle scatter with this communicator. Fall back on another component\n"));
|
|
/* HAN cannot work with this communicator so fallback on all collectives */
|
|
HAN_LOAD_FALLBACK_COLLECTIVES(han_module, comm);
|
|
return comm->c_coll->coll_scatter(sbuf, scount, sdtype, rbuf, rcount, rdtype, root,
|
|
comm, comm->c_coll->coll_scatter_module);
|
|
}
|
|
|
|
/* Topo must be initialized to know rank distribution which then is used to
|
|
* determine if han can be used */
|
|
int* topo = mca_coll_han_topo_init(comm, han_module, 2);
|
|
if (han_module->are_ppn_imbalanced) {
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output,
|
|
"han cannot handle scatter with this communicator (imbalance). Fall back on another component\n"));
|
|
/* Put back the fallback collective support and call it once. All
|
|
* future calls will then be automatically redirected.
|
|
*/
|
|
HAN_LOAD_FALLBACK_COLLECTIVE(han_module, comm, scatter);
|
|
return comm->c_coll->coll_scatter(sbuf, scount, sdtype, rbuf, rcount, rdtype, root,
|
|
comm, comm->c_coll->coll_scatter_module);
|
|
}
|
|
|
|
ompi_communicator_t *low_comm =
|
|
han_module->cached_low_comms[mca_coll_han_component.han_scatter_low_module];
|
|
ompi_communicator_t *up_comm =
|
|
han_module->cached_up_comms[mca_coll_han_component.han_scatter_up_module];
|
|
int *vranks = han_module->cached_vranks;
|
|
int low_rank = ompi_comm_rank(low_comm);
|
|
int low_size = ompi_comm_size(low_comm);
|
|
int up_size = ompi_comm_size(up_comm);
|
|
|
|
/* Set up request */
|
|
ompi_request_t *temp_request = OBJ_NEW(ompi_request_t);
|
|
temp_request->req_state = OMPI_REQUEST_ACTIVE;
|
|
temp_request->req_type = OMPI_REQUEST_COLL;
|
|
temp_request->req_free = han_request_free;
|
|
temp_request->req_status = (ompi_status_public_t){0};
|
|
temp_request->req_complete = REQUEST_PENDING;
|
|
|
|
int root_low_rank;
|
|
int root_up_rank;
|
|
|
|
mca_coll_han_get_ranks(vranks, root, low_size, &root_low_rank, &root_up_rank);
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output,
|
|
"[%d]: Han Scatter root %d root_low_rank %d root_up_rank %d\n", w_rank,
|
|
root, root_low_rank, root_up_rank));
|
|
|
|
/* Reorder sbuf based on rank.
|
|
* Suppose, message is 0 1 2 3 4 5 6 7
|
|
* and the processes are mapped on 2 nodes (the processes on the node 0 is 0 2 4 6 and the processes on the node 1 is 1 3 5 7),
|
|
* so the message needs to be reordered to 0 2 4 6 1 3 5 7
|
|
*/
|
|
char *reorder_buf = NULL;
|
|
char *reorder_sbuf = NULL;
|
|
|
|
if (w_rank == root) {
|
|
/* If the processes are mapped-by core, no need to reorder */
|
|
if (han_module->is_mapbycore) {
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output,
|
|
"[%d]: Han Scatter is_bycore: ", w_rank));
|
|
reorder_sbuf = (char *) sbuf;
|
|
} else {
|
|
ptrdiff_t ssize, sgap = 0, sextent;
|
|
ompi_datatype_type_extent(sdtype, &sextent);
|
|
ssize = opal_datatype_span(&sdtype->super, (int64_t) scount * w_size, &sgap);
|
|
reorder_buf = (char *) malloc(ssize);
|
|
reorder_sbuf = reorder_buf - sgap;
|
|
for (i = 0; i < up_size; i++) {
|
|
for (j = 0; j < low_size; j++) {
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output,
|
|
"[%d]: Han Scatter copy from %d %d\n", w_rank,
|
|
(i * low_size + j) * 2 + 1,
|
|
topo[(i * low_size + j) * 2 + 1]));
|
|
ompi_datatype_copy_content_same_ddt(sdtype, (ptrdiff_t) scount,
|
|
reorder_sbuf + sextent * (i * low_size +
|
|
j) *
|
|
(ptrdiff_t) scount,
|
|
(char *) sbuf +
|
|
sextent *
|
|
(ptrdiff_t) topo[(i * low_size + j) * 2 +
|
|
1] * (ptrdiff_t) scount);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
void *dest_buf = rbuf;
|
|
int dest_count = rcount;
|
|
ompi_datatype_t *dest_dtype = rdtype;
|
|
if (MPI_IN_PLACE == rbuf) {
|
|
dest_buf = (void*)sbuf;
|
|
dest_count = scount;
|
|
dest_dtype = sdtype;
|
|
}
|
|
|
|
/* Create us task */
|
|
mca_coll_task_t *us = OBJ_NEW(mca_coll_task_t);
|
|
/* Setup us task arguments */
|
|
mca_coll_han_scatter_args_t *us_args = malloc(sizeof(mca_coll_han_scatter_args_t));
|
|
mca_coll_han_set_scatter_args(us_args, us, reorder_sbuf, NULL, reorder_buf, scount, sdtype,
|
|
(char *) dest_buf, dest_count, dest_dtype, root, root_up_rank, root_low_rank,
|
|
up_comm, low_comm, w_rank, low_rank != root_low_rank,
|
|
temp_request);
|
|
/* Init us task */
|
|
init_task(us, mca_coll_han_scatter_us_task, (void *) (us_args));
|
|
/* Issure us task */
|
|
issue_task(us);
|
|
|
|
ompi_request_wait(&temp_request, MPI_STATUS_IGNORE);
|
|
return OMPI_SUCCESS;
|
|
|
|
}
|
|
|
|
/* us: upper level (intra-node) scatter task */
|
|
int mca_coll_han_scatter_us_task(void *task_args)
|
|
{
|
|
mca_coll_han_scatter_args_t *t = (mca_coll_han_scatter_args_t *) task_args;
|
|
|
|
if (t->noop) {
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output, "[%d] Han Scatter: us noop\n",
|
|
t->w_rank));
|
|
} else {
|
|
size_t count;
|
|
ompi_datatype_t *dtype;
|
|
if (t->w_rank == t->root) {
|
|
dtype = t->sdtype;
|
|
count = t->scount;
|
|
} else {
|
|
dtype = t->rdtype;
|
|
count = t->rcount;
|
|
}
|
|
int low_size = ompi_comm_size(t->low_comm);
|
|
ptrdiff_t rsize, rgap = 0;
|
|
rsize = opal_datatype_span(&dtype->super, (int64_t) count * low_size, &rgap);
|
|
char *tmp_buf = (char *) malloc(rsize);
|
|
char *tmp_rbuf = tmp_buf - rgap;
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output,
|
|
"[%d] Han Scatter: us scatter\n", t->w_rank));
|
|
/* Inter node scatter */
|
|
t->up_comm->c_coll->coll_scatter((char *) t->sbuf, t->scount * low_size, t->sdtype,
|
|
tmp_rbuf, t->rcount * low_size, t->rdtype, t->root_up_rank,
|
|
t->up_comm, t->up_comm->c_coll->coll_scatter_module);
|
|
t->sbuf = tmp_rbuf;
|
|
t->sbuf_inter_free = tmp_buf;
|
|
}
|
|
|
|
if (t->sbuf_reorder_free != NULL && t->root == t->w_rank) {
|
|
free(t->sbuf_reorder_free);
|
|
t->sbuf_reorder_free = NULL;
|
|
}
|
|
/* Create ls tasks for the current union segment */
|
|
mca_coll_task_t *ls = t->cur_task;
|
|
/* Init ls task */
|
|
init_task(ls, mca_coll_han_scatter_ls_task, (void *) t);
|
|
/* Issure ls task */
|
|
issue_task(ls);
|
|
|
|
return OMPI_SUCCESS;
|
|
}
|
|
|
|
/* ls: lower level (shared memory or intra-node) scatter task */
|
|
int mca_coll_han_scatter_ls_task(void *task_args)
|
|
{
|
|
mca_coll_han_scatter_args_t *t = (mca_coll_han_scatter_args_t *) task_args;
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output, "[%d] Han Scatter: ls\n",
|
|
t->w_rank));
|
|
OBJ_RELEASE(t->cur_task);
|
|
|
|
t->low_comm->c_coll->coll_scatter((char *) t->sbuf, t->scount, t->sdtype, (char *) t->rbuf,
|
|
t->rcount, t->rdtype, t->root_low_rank, t->low_comm,
|
|
t->low_comm->c_coll->coll_scatter_module);
|
|
|
|
if (t->sbuf_inter_free != NULL && t->noop != true) {
|
|
free(t->sbuf_inter_free);
|
|
t->sbuf_inter_free = NULL;
|
|
}
|
|
OPAL_OUTPUT_VERBOSE((30, mca_coll_han_component.han_output, "[%d] Han Scatter: ls finish\n",
|
|
t->w_rank));
|
|
ompi_request_t *temp_req = t->req;
|
|
free(t);
|
|
ompi_request_complete(temp_req, 1);
|
|
return OMPI_SUCCESS;
|
|
}
|