ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность

* Enable eager sending for Accumulate

Просмотр исходного кода 
* If the accumulate is local, make it short-circuit the request path.  Accumulate requires local
  ops due to its window rules, so this is likely to help a bunch (on the codes I"m messing
  with at least)
* Due a better job at flushing everything that can go out on the wire in a resource constrained problem
* Move some debugging values around to make large problems somewhat easier to deal with

This commit was SVN r20277.
Этот коммит содержится в:
Brian Barrett 2009-01-14 20:15:15 +00:00
родитель 333f419f4f
Коммит cfc400eb57
3 изменённых файлов: 101 добавлений и 27 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

41
ompi/mca/osc/rdma/osc_rdma_comm.c
Просмотреть файл

@ -26,6 +26,7 @@
#include "osc_rdma_header.h" #include "osc_rdma_header.h"
#include "osc_rdma_data_move.h" #include "osc_rdma_data_move.h"
#include "ompi/memchecker.h" #include "ompi/memchecker.h"
#include "ompi/mca/osc/base/osc_base_obj_convert.h"
static int static int
enqueue_sendreq(ompi_osc_rdma_module_t *module, enqueue_sendreq(ompi_osc_rdma_module_t *module,
@ -87,7 +88,45 @@ ompi_osc_rdma_module_accumulate(void *origin_addr, int origin_count,
sendreq->req_op_id = op->o_f_to_c_index; sendreq->req_op_id = op->o_f_to_c_index;
if (0 && module->m_eager_send_active) { if (module->m_eager_send_active) {
/* accumulate semantics require send to self, which is bloody
expensive with the extra copies. Put a shortcut in for the
common case. */
if (target == ompi_comm_rank(sendreq->req_module->m_comm) &&
ompi_ddt_is_contiguous_memory_layout(sendreq->req_target_datatype,
sendreq->req_target_count) &&
!ompi_convertor_need_buffers(&sendreq->req_origin_convertor) &&
0 == OPAL_THREAD_TRYLOCK(&module->m_acc_lock)) {
void *target_buffer = (unsigned char*) module->m_win->w_baseptr +
((unsigned long) target_disp *
module->m_win->w_disp_unit);
struct iovec iov;
uint32_t iov_count = 1;
size_t max_data = sendreq->req_origin_bytes_packed;
iov.iov_len = max_data;
iov.iov_base = NULL;
ret = ompi_convertor_pack(&sendreq->req_origin_convertor,
&iov, &iov_count,
&max_data);
if (ret < 0) {
OPAL_THREAD_UNLOCK(&module->m_acc_lock);
return OMPI_ERR_FATAL;
}
ret = ompi_osc_base_process_op(target_buffer,
iov.iov_base,
max_data,
target_dt,
target_count,
op);
/* unlock the window for accumulates */
OPAL_THREAD_UNLOCK(&module->m_acc_lock);
ompi_osc_rdma_sendreq_free(sendreq);
return ret;
}
OPAL_THREAD_LOCK(&module->m_lock); OPAL_THREAD_LOCK(&module->m_lock);
sendreq->req_module->m_num_pending_out += 1; sendreq->req_module->m_num_pending_out += 1;
module->m_num_pending_sendreqs[sendreq->req_target_rank] += 1; module->m_num_pending_sendreqs[sendreq->req_target_rank] += 1;

30
ompi/mca/osc/rdma/osc_rdma_data_move.c
Просмотреть файл

@ -372,22 +372,30 @@ ompi_osc_rdma_sendreq_send_cb(struct mca_btl_base_module_t* btl,
/* release the descriptor and sendreq */ /* release the descriptor and sendreq */
btl->btl_free(btl, descriptor); btl->btl_free(btl, descriptor);
while (opal_list_get_size(&module->m_queued_sendreqs)) { if (opal_list_get_size(&module->m_queued_sendreqs) > 0) {
opal_list_item_t *item; opal_list_item_t *item;
int ret; int ret, i, len;
OPAL_THREAD_LOCK(&module->m_lock); len = opal_list_get_size(&module->m_queued_sendreqs);
item = opal_list_remove_first(&module->m_queued_sendreqs); OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
OPAL_THREAD_UNLOCK(&module->m_lock); "%d items in restart queue",
if (NULL == item) break; len));
for (i = 0 ; i < len ; ++i) {
ret = ompi_osc_rdma_sendreq_send(module, (ompi_osc_rdma_sendreq_t*) item);
if (OMPI_SUCCESS != ret) {
OPAL_THREAD_LOCK(&module->m_lock); OPAL_THREAD_LOCK(&module->m_lock);
opal_list_append(&(module->m_queued_sendreqs), item); item = opal_list_remove_first(&module->m_queued_sendreqs);
OPAL_THREAD_UNLOCK(&module->m_lock); OPAL_THREAD_UNLOCK(&module->m_lock);
break; if (NULL == item) break;
ret = ompi_osc_rdma_sendreq_send(module, (ompi_osc_rdma_sendreq_t*) item);
if (OMPI_SUCCESS != ret) {
OPAL_THREAD_LOCK(&module->m_lock);
opal_list_append(&(module->m_queued_sendreqs), item);
OPAL_THREAD_UNLOCK(&module->m_lock);
}
} }
/* flush so things actually get sent out and resources restored */
ompi_osc_rdma_flush(module);
} }
} }

57
ompi/mca/osc/rdma/osc_rdma_sync.c
Просмотреть файл

@ -55,7 +55,7 @@ int
ompi_osc_rdma_module_fence(int assert, ompi_win_t *win) ompi_osc_rdma_module_fence(int assert, ompi_win_t *win)
{ {
unsigned int incoming_reqs; unsigned int incoming_reqs;
int ret = OMPI_SUCCESS, i; int ret = OMPI_SUCCESS, i, len, started_send;
ompi_osc_rdma_module_t *module = GET_MODULE(win); ompi_osc_rdma_module_t *module = GET_MODULE(win);
int num_outgoing = 0; int num_outgoing = 0;
@ -112,18 +112,45 @@ ompi_osc_rdma_module_fence(int assert, ompi_win_t *win)
/* try to start all the requests. We've copied everything we /* try to start all the requests. We've copied everything we
need out of pending_sendreqs, so don't need the lock need out of pending_sendreqs, so don't need the lock
here */ here */
while (NULL != len = opal_list_get_size(&(module->m_copy_pending_sendreqs));
(item = opal_list_remove_first(&(module->m_copy_pending_sendreqs)))) { started_send = 0;
ompi_osc_rdma_sendreq_t *req = OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
(ompi_osc_rdma_sendreq_t*) item; "fence: trying to start %d reqs",
len));
for (i = 0 ; i < len ; ++i) {
ompi_osc_rdma_sendreq_t *req = (ompi_osc_rdma_sendreq_t*)
opal_list_remove_first(&(module->m_copy_pending_sendreqs));
ret = ompi_osc_rdma_sendreq_send(module, req); ret = ompi_osc_rdma_sendreq_send(module, req);
if (OMPI_SUCCESS != ret) { if (OMPI_SUCCESS != ret) {
opal_list_append(&(module->m_copy_pending_sendreqs), item); opal_list_append(&(module->m_copy_pending_sendreqs), item);
break; } else {
started_send = 1;
} }
} }
/* we need to start at least one send, so that the callback
will restart the rest. */
while (0 == started_send && len != 0) {
opal_progress();
OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"fence: restarting %d reqs", len));
len = opal_list_get_size(&(module->m_copy_pending_sendreqs));
for (i = 0 ; i < len ; ++i) {
ompi_osc_rdma_sendreq_t *req = (ompi_osc_rdma_sendreq_t*)
opal_list_remove_first(&(module->m_copy_pending_sendreqs));
ret = ompi_osc_rdma_sendreq_send(module, req);
if (OMPI_SUCCESS != ret) {
opal_list_append(&(module->m_copy_pending_sendreqs), item);
} else {
started_send = 1;
}
}
}
OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"fence: done with initial start"));
if (module->m_use_rdma) { if (module->m_use_rdma) {
if (module->m_rdma_wait_completion) { if (module->m_rdma_wait_completion) {
OPAL_THREAD_LOCK(&module->m_lock); OPAL_THREAD_LOCK(&module->m_lock);
@ -167,7 +194,7 @@ ompi_osc_rdma_module_fence(int assert, ompi_win_t *win)
module->m_num_pending_in += incoming_reqs; module->m_num_pending_in += incoming_reqs;
module->m_num_pending_out += num_outgoing; module->m_num_pending_out += num_outgoing;
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"fence: waiting on %d in and %d out, now %d, %d", "fence: waiting on %d in and %d out, now %d, %d",
incoming_reqs, incoming_reqs,
num_outgoing, num_outgoing,
@ -484,7 +511,7 @@ ompi_osc_rdma_module_lock(int lock_type,
ompi_win_remove_mode(win, OMPI_WIN_FENCE); ompi_win_remove_mode(win, OMPI_WIN_FENCE);
ompi_win_append_mode(win, OMPI_WIN_ACCESS_EPOCH | OMPI_WIN_LOCK_ACCESS); ompi_win_append_mode(win, OMPI_WIN_ACCESS_EPOCH | OMPI_WIN_LOCK_ACCESS);
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"%d sending lock request to %d", "%d sending lock request to %d",
ompi_comm_rank(module->m_comm), target)); ompi_comm_rank(module->m_comm), target));
/* generate a lock request */ /* generate a lock request */
@ -532,7 +559,7 @@ ompi_osc_rdma_module_unlock(int target,
OPAL_THREAD_UNLOCK(&module->m_lock); OPAL_THREAD_UNLOCK(&module->m_lock);
/* send the unlock request */ /* send the unlock request */
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"%d sending unlock request to %d with %d requests", "%d sending unlock request to %d with %d requests",
ompi_comm_rank(module->m_comm), target, ompi_comm_rank(module->m_comm), target,
out_count)); out_count));
@ -600,7 +627,7 @@ ompi_osc_rdma_passive_lock(ompi_osc_rdma_module_t *module,
ompi_win_append_mode(module->m_win, OMPI_WIN_EXPOSE_EPOCH); ompi_win_append_mode(module->m_win, OMPI_WIN_EXPOSE_EPOCH);
send_ack = true; send_ack = true;
} else { } else {
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"%d queuing lock request from %d (%d)", "%d queuing lock request from %d (%d)",
ompi_comm_rank(module->m_comm), ompi_comm_rank(module->m_comm),
origin, lock_type)); origin, lock_type));
@ -616,7 +643,7 @@ ompi_osc_rdma_passive_lock(ompi_osc_rdma_module_t *module,
ompi_win_append_mode(module->m_win, OMPI_WIN_EXPOSE_EPOCH); ompi_win_append_mode(module->m_win, OMPI_WIN_EXPOSE_EPOCH);
send_ack = true; send_ack = true;
} else { } else {
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"queuing lock request from %d (%d) lock_type:%d", "queuing lock request from %d (%d) lock_type:%d",
ompi_comm_rank(module->m_comm), ompi_comm_rank(module->m_comm),
origin, lock_type)); origin, lock_type));
@ -631,7 +658,7 @@ ompi_osc_rdma_passive_lock(ompi_osc_rdma_module_t *module,
OPAL_THREAD_UNLOCK(&(module->m_lock)); OPAL_THREAD_UNLOCK(&(module->m_lock));
if (send_ack) { if (send_ack) {
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"%d sending lock ack to %d", "%d sending lock ack to %d",
ompi_comm_rank(module->m_comm), origin)); ompi_comm_rank(module->m_comm), origin));
ompi_osc_rdma_control_send(module, proc, ompi_osc_rdma_control_send(module, proc,
@ -654,7 +681,7 @@ ompi_osc_rdma_passive_unlock(ompi_osc_rdma_module_t *module,
assert(module->m_lock_status != 0); assert(module->m_lock_status != 0);
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"received unlock request from %d with %d requests\n", "received unlock request from %d with %d requests\n",
origin, count)); origin, count));
@ -706,7 +733,7 @@ ompi_osc_rdma_passive_unlock_complete(ompi_osc_rdma_module_t *module)
/* issue whichever unlock acks we should issue */ /* issue whichever unlock acks we should issue */
while (NULL != (new_pending = (ompi_osc_rdma_pending_lock_t*) while (NULL != (new_pending = (ompi_osc_rdma_pending_lock_t*)
opal_list_remove_first(&copy_unlock_acks))) { opal_list_remove_first(&copy_unlock_acks))) {
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"sending unlock reply to proc")); "sending unlock reply to proc"));
ompi_osc_rdma_control_send(module, ompi_osc_rdma_control_send(module,
new_pending->proc, new_pending->proc,
@ -735,7 +762,7 @@ ompi_osc_rdma_passive_unlock_complete(ompi_osc_rdma_module_t *module)
OPAL_THREAD_UNLOCK(&(module->m_lock)); OPAL_THREAD_UNLOCK(&(module->m_lock));
if (NULL != new_pending) { if (NULL != new_pending) {
OPAL_OUTPUT_VERBOSE((50, ompi_osc_base_output, OPAL_OUTPUT_VERBOSE((40, ompi_osc_base_output,
"sending lock request to proc")); "sending lock request to proc"));
ompi_osc_rdma_control_send(module, ompi_osc_rdma_control_send(module,
new_pending->proc, new_pending->proc,