MPI_Win smpi_mpi_win_create( void *base, MPI_Aint size, int disp_unit, MPI_Info info, MPI_Comm comm){
-
MPI_Win win;
-
+
int comm_size = smpi_comm_size(comm);
int rank=smpi_comm_rank(comm);
XBT_DEBUG("Creating window");
win->requests = xbt_dynar_new(sizeof(MPI_Request), NULL);
win->connected_wins = xbt_new0(MPI_Win, comm_size);
win->connected_wins[rank] = win;
-
+
if(rank==0){
win->bar=xbt_barrier_init(comm_size);
}
-
- mpi_coll_allgather_fun(&(win->connected_wins[rank]),
- sizeof(MPI_Win),
- MPI_BYTE,
- win->connected_wins,
- sizeof(MPI_Win),
- MPI_BYTE,
- comm);
-
- mpi_coll_bcast_fun( &(win->bar),
- sizeof(xbt_bar_t),
- MPI_BYTE,
- 0,
- comm);
-
+
+ mpi_coll_allgather_fun(&(win->connected_wins[rank]), sizeof(MPI_Win), MPI_BYTE, win->connected_wins, sizeof(MPI_Win),
+ MPI_BYTE, comm);
+
+ mpi_coll_bcast_fun( &(win->bar), sizeof(xbt_bar_t), MPI_BYTE, 0, comm);
+
mpi_coll_barrier_fun(comm);
-
+
return win;
}
int smpi_mpi_win_free( MPI_Win* win){
-
//As per the standard, perform a barrier to ensure every async comm is finished
xbt_barrier_wait((*win)->bar);
xbt_dynar_free(&(*win)->requests);
win->name = xbt_strdup(name);;
}
-
int smpi_mpi_win_fence( int assert, MPI_Win win){
-
XBT_DEBUG("Entering fence");
if(!win->opened)
win->opened=1;
}
MPI_Request* treqs = static_cast<MPI_Request*>(xbt_dynar_to_array(reqs));
+ win->requests=xbt_dynar_new(sizeof(MPI_Request), NULL);
smpi_mpi_waitall(size,treqs,MPI_STATUSES_IGNORE);
xbt_free(treqs);
- win->requests=xbt_dynar_new(sizeof(MPI_Request), NULL);
}
win->assert = assert;
-
+
xbt_barrier_wait(win->bar);
XBT_DEBUG("Leaving fence ");
int smpi_mpi_put( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, int target_rank,
MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Win win)
{
-
if(!win->opened)//check that post/start has been done
return MPI_ERR_WIN;
//get receiver pointer
if(target_rank != smpi_comm_rank(win->comm)){
//prepare send_request
- MPI_Request sreq = smpi_rma_send_init(origin_addr, origin_count, origin_datatype,
- smpi_process_index(), smpi_group_index(smpi_comm_group(win->comm),target_rank), RMA_TAG+1, win->comm, MPI_OP_NULL);
+ MPI_Request sreq = smpi_rma_send_init(origin_addr, origin_count, origin_datatype, smpi_process_index(),
+ smpi_group_index(smpi_comm_group(win->comm),target_rank), RMA_TAG+1, win->comm, MPI_OP_NULL);
//prepare receiver request
- MPI_Request rreq = smpi_rma_recv_init(recv_addr, target_count, target_datatype,
- smpi_process_index(), smpi_group_index(smpi_comm_group(win->comm),target_rank), RMA_TAG+1, recv_win->comm, MPI_OP_NULL);
+ MPI_Request rreq = smpi_rma_recv_init(recv_addr, target_count, target_datatype, smpi_process_index(),
+ smpi_group_index(smpi_comm_group(win->comm),target_rank), RMA_TAG+1, recv_win->comm, MPI_OP_NULL);
//push request to receiver's win
xbt_dynar_push_as(recv_win->requests, MPI_Request, rreq);
//push request to sender's win
xbt_dynar_push_as(win->requests, MPI_Request, sreq);
}else{
- smpi_datatype_copy(origin_addr, origin_count, origin_datatype,
- recv_addr, target_count, target_datatype);
+ smpi_datatype_copy(origin_addr, origin_count, origin_datatype, recv_addr, target_count, target_datatype);
}
return MPI_SUCCESS;
if(target_rank != smpi_comm_rank(win->comm)){
//prepare send_request
MPI_Request sreq = smpi_rma_send_init(send_addr, target_count, target_datatype,
- smpi_group_index(smpi_comm_group(win->comm),target_rank), smpi_process_index(), RMA_TAG+2, send_win->comm, MPI_OP_NULL);
+ smpi_group_index(smpi_comm_group(win->comm),target_rank), smpi_process_index(), RMA_TAG+2, send_win->comm,
+ MPI_OP_NULL);
//prepare receiver request
MPI_Request rreq = smpi_rma_recv_init(origin_addr, origin_count, origin_datatype,
- smpi_group_index(smpi_comm_group(win->comm),target_rank), smpi_process_index(), RMA_TAG+2, win->comm, MPI_OP_NULL);
-
+ smpi_group_index(smpi_comm_group(win->comm),target_rank), smpi_process_index(), RMA_TAG+2, win->comm,
+ MPI_OP_NULL);
+
//start the send, with another process than us as sender.
smpi_mpi_start(sreq);
-
+
//push request to receiver's win
xbt_dynar_push_as(send_win->requests, MPI_Request, sreq);
//push request to sender's win
xbt_dynar_push_as(win->requests, MPI_Request, rreq);
}else{
- smpi_datatype_copy(send_addr, target_count, target_datatype,
- origin_addr, origin_count, origin_datatype);
+ smpi_datatype_copy(send_addr, target_count, target_datatype, origin_addr, origin_count, origin_datatype);
}
return MPI_SUCCESS;
smpi_datatype_use(origin_datatype);
smpi_datatype_use(target_datatype);
-
//prepare send_request
MPI_Request sreq = smpi_rma_send_init(origin_addr, origin_count, origin_datatype,
smpi_process_index(), smpi_group_index(smpi_comm_group(win->comm),target_rank), RMA_TAG+3, win->comm, op);
xbt_dynar_push_as(recv_win->requests, MPI_Request, rreq);
//start send
smpi_mpi_start(sreq);
-
+
//push request to sender's win
xbt_dynar_push_as(win->requests, MPI_Request, sreq);
-
-
return MPI_SUCCESS;
}
int smpi_mpi_win_start(MPI_Group group, int assert, MPI_Win win){
/* From MPI forum advices
- The call to MPI_WIN_COMPLETE does not return until the put call has completed at
- the origin; and the target window will be accessed by the put operation only
- after the call to MPI_WIN_START has matched a call to MPI_WIN_POST by the target
- process. This still leaves much choice to implementors. The call to
- MPI_WIN_START can block until the matching call to MPI_WIN_POST occurs at all
- target processes. One can also have implementations where the call to
- MPI_WIN_START is nonblocking, but the call to MPI_PUT blocks until the matching
- call to MPI_WIN_POST occurred; or implementations where the first two calls are
- nonblocking, but the call to MPI_WIN_COMPLETE blocks until the call to
- MPI_WIN_POST occurred; or even implementations where all three calls can
- complete before any target process called MPI_WIN_POST --- the data put must be
- buffered, in this last case, so as to allow the put to complete at the origin
- ahead of its completion at the target. However, once the call to MPI_WIN_POST is
- issued, the sequence above must complete, without further dependencies.
- */
-
+ The call to MPI_WIN_COMPLETE does not return until the put call has completed at the origin; and the target window
+ will be accessed by the put operation only after the call to MPI_WIN_START has matched a call to MPI_WIN_POST by
+ the target process. This still leaves much choice to implementors. The call to MPI_WIN_START can block until the
+ matching call to MPI_WIN_POST occurs at all target processes. One can also have implementations where the call to
+ MPI_WIN_START is nonblocking, but the call to MPI_PUT blocks until the matching call to MPI_WIN_POST occurred; or
+ implementations where the first two calls are nonblocking, but the call to MPI_WIN_COMPLETE blocks until the call
+ to MPI_WIN_POST occurred; or even implementations where all three calls can complete before any target process
+ called MPI_WIN_POST --- the data put must be buffered, in this last case, so as to allow the put to complete at the
+ origin ahead of its completion at the target. However, once the call to MPI_WIN_POST is issued, the sequence above
+ must complete, without further dependencies. */
+
//naive, blocking implementation.
int i=0,j=0;
int size = smpi_group_size(group);
MPI_Request* reqs = xbt_new0(MPI_Request, size);
-
+
// for(i=0;i<size;i++){
while(j!=size){
int src=smpi_group_index(group,j);
int i=0,j=0;
int size = smpi_group_size(group);
MPI_Request* reqs = xbt_new0(MPI_Request, size);
-
+
while(j!=size){
int dst=smpi_group_index(group,j);
if(dst!=smpi_process_index()){
- reqs[i]=smpi_mpi_send_init(NULL, 0, MPI_CHAR, dst,
- RMA_TAG+4, MPI_COMM_WORLD);
+ reqs[i]=smpi_mpi_send_init(NULL, 0, MPI_CHAR, dst, RMA_TAG+4, MPI_COMM_WORLD);
i++;
}
j++;
}
size=i;
-
+
smpi_mpi_startall(size, reqs);
smpi_mpi_waitall(size, reqs, MPI_STATUSES_IGNORE);
for(i=0;i<size;i++){
//MPI_Comm comm = smpi_comm_new(win->group, NULL);
//mpi_coll_barrier_fun(comm);
//smpi_comm_destroy(comm);
-
+
XBT_DEBUG("Entering MPI_Win_Complete");
int i=0,j=0;
int size = smpi_group_size(win->group);
MPI_Request* reqs = xbt_new0(MPI_Request, size);
-
+
while(j!=size){
int dst=smpi_group_index(win->group,j);
if(dst!=smpi_process_index()){
- reqs[i]=smpi_mpi_send_init(NULL, 0, MPI_CHAR, dst,
- RMA_TAG+5, MPI_COMM_WORLD);
+ reqs[i]=smpi_mpi_send_init(NULL, 0, MPI_CHAR, dst, RMA_TAG+5, MPI_COMM_WORLD);
i++;
}
j++;
XBT_DEBUG("Win_complete - Sending sync messages to %d processes", size);
smpi_mpi_startall(size, reqs);
smpi_mpi_waitall(size, reqs, MPI_STATUSES_IGNORE);
-
+
for(i=0;i<size;i++){
smpi_mpi_request_free(&reqs[i]);
}
xbt_free(reqs);
-
+
//now we can finish RMA calls
-
+
xbt_dynar_t reqqs = win->requests;
size = xbt_dynar_length(reqqs);
-
+
XBT_DEBUG("Win_complete - Finishing %d RMA calls", size);
unsigned int cpt=0;
MPI_Request req;
}
MPI_Request* treqs = static_cast<MPI_Request*>(xbt_dynar_to_array(reqqs));
+ win->requests=xbt_dynar_new(sizeof(MPI_Request), NULL);
smpi_mpi_waitall(size,treqs,MPI_STATUSES_IGNORE);
xbt_free(treqs);
- win->requests=xbt_dynar_new(sizeof(MPI_Request), NULL);
win->opened--; //we're closed for business !
return MPI_SUCCESS;
}
-
-
int smpi_mpi_win_wait(MPI_Win win){
// xbt_barrier_wait(win->bar);
//MPI_Comm comm = smpi_comm_new(win->group, NULL);
int i=0,j=0;
int size = smpi_group_size(win->group);
MPI_Request* reqs = xbt_new0(MPI_Request, size);
-
+
// for(i=0;i<size;i++){
while(j!=size){
int src=smpi_group_index(win->group,j);
xbt_dynar_t reqqs = win->requests;
size = xbt_dynar_length(reqqs);
-
+
XBT_DEBUG("Win_complete - Finishing %d RMA calls", size);
unsigned int cpt=0;
}
MPI_Request* treqs = static_cast<MPI_Request*>(xbt_dynar_to_array(reqqs));
+ win->requests=xbt_dynar_new(sizeof(MPI_Request), NULL);
smpi_mpi_waitall(size,treqs,MPI_STATUSES_IGNORE);
xbt_free(treqs);
- win->requests=xbt_dynar_new(sizeof(MPI_Request), NULL);
win->opened--; //we're opened for business !
return MPI_SUCCESS;
}