Skip to content

GitLab

Commit f76f53ae authored by James Dinan's avatar James Dinan
Browse files

[svn-r10683] RMA Req-ops: Improved self and PROC_NULL handling 

This patch more efficiently handles request-generating operations that target
the local process and MPI_PROC_NULL.  These special cases return completed
generalized requests, which cost less overhead.  This also makes space for a
more efficient implementation when direct, shared memory operations are
implemented.

Reviewer: goodell
parent 81dfdbea
Hide whitespace changes
Inline Side-by-side
src/mpid/ch3/src/ch3u_rma_reqops.c
View file @ f76f53ae
......@@ -30,19 +30,12 @@ static int MPIDI_CH3I_Rma_req_poll(void *state, MPI_Status *status)
MPIU_UNREFERENCED_ARG(status);
/* If this is a local operation, it's already complete. Otherwise, call
* flush to complete the operation */
/* FIXME: We still may need to flush or sync for shared memory windows */
if (req_state->target_rank != req_state->win_ptr->comm_ptr->rank &&
req_state->target_rank != MPI_PROC_NULL)
{
mpi_errno = req_state->win_ptr->RMAFns.Win_flush(req_state->target_rank,
req_state->win_ptr);
if (mpi_errno != MPI_SUCCESS) {
MPIU_ERR_SETANDJUMP(mpi_errno, MPI_ERR_OTHER, "**rmareqop");
}
}
/* Call flush to complete the operation */
/* FIXME: We need per-operation completion to make this more efficient. */
mpi_errno = req_state->win_ptr->RMAFns.Win_flush(req_state->target_rank,
req_state->win_ptr);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
MPIR_Grequest_complete_impl(req_state->request);
......@@ -161,22 +154,38 @@ int MPIDI_Rput(const void *origin_addr, int origin_count,
req_state->win_ptr = win_ptr;
req_state->target_rank = target_rank;
/* Enqueue the RMA operation */
mpi_errno = win_ptr->RMAFns.Put(origin_addr, origin_count,
origin_datatype, target_rank,
target_disp, target_count,
target_datatype, win_ptr);
/* Enqueue or perform the RMA operation */
if (target_rank != MPI_PROC_NULL) {
mpi_errno = win_ptr->RMAFns.Put(origin_addr, origin_count,
origin_datatype, target_rank,
target_disp, target_count,
target_datatype, win_ptr);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
/* If the operation is already complete, return a completed request.
* Otherwise, generate a grequest. */
/* FIXME: We still may need to flush or sync for shared memory windows */
if (target_rank == MPI_PROC_NULL || target_rank == win_ptr->myrank) {
mpi_errno = MPIR_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
MPIR_Grequest_complete_impl(req_state->request);
}
else {
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
*request = req_state->request;
......@@ -216,22 +225,38 @@ int MPIDI_Rget(void *origin_addr, int origin_count,
req_state->win_ptr = win_ptr;
req_state->target_rank = target_rank;
/* Enqueue the RMA operation */
mpi_errno = win_ptr->RMAFns.Get(origin_addr, origin_count,
origin_datatype, target_rank,
target_disp, target_count,
target_datatype, win_ptr);
/* Enqueue or perform the RMA operation */
if (target_rank != MPI_PROC_NULL) {
mpi_errno = win_ptr->RMAFns.Get(origin_addr, origin_count,
origin_datatype, target_rank,
target_disp, target_count,
target_datatype, win_ptr);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
/* If the operation is already complete, return a completed request.
* Otherwise, generate a grequest. */
/* FIXME: We still may need to flush or sync for shared memory windows */
if (target_rank == MPI_PROC_NULL || target_rank == win_ptr->myrank) {
mpi_errno = MPIR_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
MPIR_Grequest_complete_impl(req_state->request);
}
else {
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
*request = req_state->request;
......@@ -271,22 +296,37 @@ int MPIDI_Raccumulate(const void *origin_addr, int origin_count,
req_state->win_ptr = win_ptr;
req_state->target_rank = target_rank;
/* Enqueue the RMA operation */
mpi_errno = win_ptr->RMAFns.Accumulate(origin_addr, origin_count,
origin_datatype, target_rank,
target_disp, target_count,
target_datatype, op, win_ptr);
/* Enqueue or perform the RMA operation */
if (target_rank != MPI_PROC_NULL) {
mpi_errno = win_ptr->RMAFns.Accumulate(origin_addr, origin_count,
origin_datatype, target_rank,
target_disp, target_count,
target_datatype, op, win_ptr);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
/* If the operation is already complete, return a completed request.
* Otherwise, generate a grequest. */
/* FIXME: We still may need to flush or sync for shared memory windows */
if (target_rank == MPI_PROC_NULL || target_rank == win_ptr->myrank) {
mpi_errno = MPIR_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
MPIR_Grequest_complete_impl(req_state->request);
}
else {
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
*request = req_state->request;
......@@ -327,24 +367,39 @@ int MPIDI_Rget_accumulate(const void *origin_addr, int origin_count,
req_state->win_ptr = win_ptr;
req_state->target_rank = target_rank;
/* Enqueue the RMA operation */
mpi_errno = win_ptr->RMAFns.Get_accumulate(origin_addr, origin_count,
origin_datatype, result_addr,
result_count, result_datatype,
target_rank, target_disp,
target_count, target_datatype,
op, win_ptr);
/* Enqueue or perform the RMA operation */
if (target_rank != MPI_PROC_NULL) {
mpi_errno = win_ptr->RMAFns.Get_accumulate(origin_addr, origin_count,
origin_datatype, result_addr,
result_count, result_datatype,
target_rank, target_disp,
target_count, target_datatype,
op, win_ptr);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
/* If the operation is already complete, return a completed request.
* Otherwise, generate a grequest. */
/* FIXME: We still may need to flush or sync for shared memory windows */
if (target_rank == MPI_PROC_NULL || target_rank == win_ptr->myrank) {
mpi_errno = MPIR_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
MPIR_Grequest_complete_impl(req_state->request);
}
else {
mpi_errno = MPIX_Grequest_start_impl(MPIDI_CH3I_Rma_req_query,
MPIDI_CH3I_Rma_req_free,
MPIDI_CH3I_Rma_req_cancel,
MPIDI_CH3I_Rma_req_poll,
MPIDI_CH3I_Rma_req_wait,
req_state, &req_state->request);
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); }
}
*request = req_state->request;
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment