int ompi_coll_tuned_reduce_intra_binary( void *sendbuf, void *recvbuf,
int count, ompi_datatype_t* datatype,
ompi_op_t* op, int root,
ompi_communicator_t* comm,
mca_coll_base_module_t *module,
uint32_t segsize,
int max_outstanding_reqs )
{
int segcount = count;
size_t typelng;
mca_coll_tuned_module_t *tuned_module = (mca_coll_tuned_module_t*) module;
mca_coll_tuned_comm_t *data = tuned_module->tuned_data;
OPAL_OUTPUT((ompi_coll_tuned_stream,"coll:tuned:reduce_intra_binary rank %d ss %5d",
ompi_comm_rank(comm), segsize));
COLL_TUNED_UPDATE_BINTREE( comm, tuned_module, root );
/**
* Determine number of segments and number of elements
* sent per operation
*/
ompi_datatype_type_size( datatype, &typelng );
COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
return ompi_coll_tuned_reduce_generic( sendbuf, recvbuf, count, datatype,
op, root, comm, module,
data->cached_bintree,
segcount, max_outstanding_reqs );
}
int
ompi_coll_tuned_bcast_intra_bintree ( void* buffer,
int count,
struct ompi_datatype_t* datatype,
int root,
struct ompi_communicator_t* comm,
mca_coll_base_module_t *module,
uint32_t segsize )
{
int segcount = count;
size_t typelng;
mca_coll_tuned_module_t *tuned_module = (mca_coll_tuned_module_t*) module;
mca_coll_tuned_comm_t *data = tuned_module->tuned_data;
COLL_TUNED_UPDATE_BINTREE( comm, tuned_module, root );
/**
* Determine number of elements sent per operation.
*/
ompi_datatype_type_size( datatype, &typelng );
COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
OPAL_OUTPUT((ompi_coll_tuned_stream,"coll:tuned:bcast_intra_binary rank %d ss %5d typelng %lu segcount %d",
ompi_comm_rank(comm), segsize, (unsigned long)typelng, segcount));
return ompi_coll_tuned_bcast_intra_generic( buffer, count, datatype, root, comm, module,
segcount, data->cached_bintree );
}
int ompi_coll_tuned_reduce_intra_binary( void *sendbuf, void *recvbuf,
int count, ompi_datatype_t* datatype,
ompi_op_t* op, int root,
ompi_communicator_t* comm, uint32_t segsize )
{
int segcount = count;
size_t typelng;
OPAL_OUTPUT((ompi_coll_tuned_stream,"coll:tuned:reduce_intra_binary rank %d ss %5d",
ompi_comm_rank(comm), segsize));
COLL_TUNED_UPDATE_BINTREE( comm, root );
/**
* Determine number of segments and number of elements
* sent per operation
*/
ompi_ddt_type_size( datatype, &typelng );
COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
return ompi_coll_tuned_reduce_generic( sendbuf, recvbuf, count, datatype, op, root, comm,
comm->c_coll_selected_data->cached_bintree, segcount );
}
int
ompi_coll_tuned_bcast_intra_split_bintree ( void* buffer,
int count,
struct ompi_datatype_t* datatype,
int root,
struct ompi_communicator_t* comm,
mca_coll_base_module_t *module,
uint32_t segsize )
{
int err=0, line, rank, size, segindex, i, lr, pair;
uint32_t counts[2];
int segcount[2]; /* Number of elements sent with each segment */
int num_segments[2]; /* Number of segmenets */
int sendcount[2]; /* the same like segcount, except for the last segment */
size_t realsegsize[2], type_size;
char *tmpbuf[2];
ptrdiff_t type_extent, lb;
ompi_request_t *base_req, *new_req;
ompi_coll_tree_t *tree;
mca_coll_tuned_module_t *tuned_module = (mca_coll_tuned_module_t*) module;
mca_coll_tuned_comm_t *data = tuned_module->tuned_data;
size = ompi_comm_size(comm);
rank = ompi_comm_rank(comm);
OPAL_OUTPUT((ompi_coll_tuned_stream,"ompi_coll_tuned_bcast_intra_split_bintree rank %d root %d ss %5d", rank, root, segsize));
if (size == 1) {
return MPI_SUCCESS;
}
/* setup the binary tree topology. */
COLL_TUNED_UPDATE_BINTREE( comm, tuned_module, root );
tree = data->cached_bintree;
err = ompi_datatype_type_size( datatype, &type_size );
/* Determine number of segments and number of elements per segment */
counts[0] = count/2;
if (count % 2 != 0) counts[0]++;
counts[1] = count - counts[0];
if ( segsize > 0 ) {
/* Note that ompi_datatype_type_size() will never return a negative
value in typelng; it returns an int [vs. an unsigned type]
because of the MPI spec. */
if (segsize < ((uint32_t) type_size)) {
segsize = type_size; /* push segsize up to hold one type */
}
segcount[0] = segcount[1] = segsize / type_size;
num_segments[0] = counts[0]/segcount[0];
if ((counts[0] % segcount[0]) != 0) num_segments[0]++;
num_segments[1] = counts[1]/segcount[1];
if ((counts[1] % segcount[1]) != 0) num_segments[1]++;
} else {
segcount[0] = counts[0];
segcount[1] = counts[1];
num_segments[0] = num_segments[1] = 1;
}
/* if the message is too small to be split into segments */
if( (counts[0] == 0 || counts[1] == 0) ||
(segsize > ((ptrdiff_t)counts[0] * type_size)) ||
(segsize > ((ptrdiff_t)counts[1] * type_size)) ) {
/* call linear version here ! */
return (ompi_coll_tuned_bcast_intra_chain ( buffer, count, datatype,
root, comm, module,
segsize, 1 ));
}
err = ompi_datatype_get_extent (datatype, &lb, &type_extent);
/* Determine real segment size */
realsegsize[0] = (ptrdiff_t)segcount[0] * type_extent;
realsegsize[1] = (ptrdiff_t)segcount[1] * type_extent;
/* set the buffer pointers */
tmpbuf[0] = (char *) buffer;
tmpbuf[1] = (char *) buffer + (ptrdiff_t)counts[0] * type_extent;
/* Step 1:
Root splits the buffer in 2 and sends segmented message down the branches.
Left subtree of the tree receives first half of the buffer, while right
subtree receives the remaining message.
*/
/* determine if I am left (0) or right (1), (root is right) */
lr = ((rank + size - root)%size + 1)%2;
/* root code */
if( rank == root ) {
/* determine segment count */
sendcount[0] = segcount[0];
sendcount[1] = segcount[1];
/* for each segment */
for (segindex = 0; segindex < num_segments[0]; segindex++) {
/* for each child */
for( i = 0; i < tree->tree_nextsize && i < 2; i++ ) {
if (segindex >= num_segments[i]) { /* no more segments */
continue;
}
/* determine how many elements are being sent in this round */
if(segindex == (num_segments[i] - 1))
sendcount[i] = counts[i] - segindex*segcount[i];
/* send data */
MCA_PML_CALL(send(tmpbuf[i], sendcount[i], datatype,
tree->tree_next[i], MCA_COLL_BASE_TAG_BCAST,
MCA_PML_BASE_SEND_STANDARD, comm));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
/* update tmp buffer */
tmpbuf[i] += realsegsize[i];
}
}
}
/* intermediate nodes code */
else if( tree->tree_nextsize > 0 ) {
/* Intermediate nodes:
* It will receive segments only from one half of the data.
* Which one is determined by whether the node belongs to the "left" or "right"
* subtree. Topoloby building function builds binary tree such that
* odd "shifted ranks" ((rank + size - root)%size) are on the left subtree,
* and even on the right subtree.
*
* Create the pipeline. We first post the first receive, then in the loop we
* post the next receive and after that wait for the previous receive to complete
* and we disseminating the data to all children.
*/
sendcount[lr] = segcount[lr];
err = MCA_PML_CALL(irecv(tmpbuf[lr], sendcount[lr], datatype,
tree->tree_prev, MCA_COLL_BASE_TAG_BCAST,
comm, &base_req));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
for( segindex = 1; segindex < num_segments[lr]; segindex++ ) {
/* determine how many elements to expect in this round */
if( segindex == (num_segments[lr] - 1))
sendcount[lr] = counts[lr] - (ptrdiff_t)segindex * (ptrdiff_t)segcount[lr];
/* post new irecv */
err = MCA_PML_CALL(irecv( tmpbuf[lr] + realsegsize[lr], sendcount[lr],
datatype, tree->tree_prev, MCA_COLL_BASE_TAG_BCAST,
comm, &new_req));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
/* wait for and forward current segment */
err = ompi_request_wait_all( 1, &base_req, MPI_STATUSES_IGNORE );
for( i = 0; i < tree->tree_nextsize; i++ ) { /* send data to children (segcount[lr]) */
err = MCA_PML_CALL(send( tmpbuf[lr], segcount[lr], datatype,
tree->tree_next[i], MCA_COLL_BASE_TAG_BCAST,
MCA_PML_BASE_SEND_STANDARD, comm));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
} /* end of for each child */
/* upate the base request */
base_req = new_req;
/* go to the next buffer (ie. the one corresponding to the next recv) */
tmpbuf[lr] += realsegsize[lr];
} /* end of for segindex */
/* wait for the last segment and forward current segment */
err = ompi_request_wait_all( 1, &base_req, MPI_STATUSES_IGNORE );
for( i = 0; i < tree->tree_nextsize; i++ ) { /* send data to children */
err = MCA_PML_CALL(send(tmpbuf[lr], sendcount[lr], datatype,
tree->tree_next[i], MCA_COLL_BASE_TAG_BCAST,
MCA_PML_BASE_SEND_STANDARD, comm));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
} /* end of for each child */
}
/* leaf nodes */
else {
/* Just consume segments as fast as possible */
sendcount[lr] = segcount[lr];
for (segindex = 0; segindex < num_segments[lr]; segindex++) {
/* determine how many elements to expect in this round */
if (segindex == (num_segments[lr] - 1))
sendcount[lr] = counts[lr] - (ptrdiff_t)segindex * (ptrdiff_t)segcount[lr];
/* receive segments */
err = MCA_PML_CALL(recv(tmpbuf[lr], sendcount[lr], datatype,
tree->tree_prev, MCA_COLL_BASE_TAG_BCAST,
comm, MPI_STATUS_IGNORE));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
/* update the initial pointer to the buffer */
tmpbuf[lr] += realsegsize[lr];
}
}
/* reset the buffer pointers */
tmpbuf[0] = (char *) buffer;
tmpbuf[1] = (char *) buffer + (ptrdiff_t)counts[0] * type_extent;
/* Step 2:
Find your immediate pair (identical node in opposite subtree) and SendRecv
data buffer with them.
The tree building function ensures that
if (we are not root)
if we are in the left subtree (lr == 0) our pair is (rank+1)%size.
if we are in the right subtree (lr == 1) our pair is (rank-1)%size
If we have even number of nodes the rank (size-1) will pair up with root.
*/
if (lr == 0) {
pair = (rank+1)%size;
} else {
pair = (rank+size-1)%size;
}
if ( (size%2) != 0 && rank != root) {
err = ompi_coll_tuned_sendrecv( tmpbuf[lr], counts[lr], datatype,
pair, MCA_COLL_BASE_TAG_BCAST,
tmpbuf[(lr+1)%2], counts[(lr+1)%2], datatype,
pair, MCA_COLL_BASE_TAG_BCAST,
comm, MPI_STATUS_IGNORE, rank);
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
} else if ( (size%2) == 0 ) {
/* root sends right buffer to the last node */
if( rank == root ) {
err = MCA_PML_CALL(send(tmpbuf[1], counts[1], datatype,
(root+size-1)%size, MCA_COLL_BASE_TAG_BCAST,
MCA_PML_BASE_SEND_STANDARD, comm));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
}
/* last node receives right buffer from the root */
else if (rank == (root+size-1)%size) {
err = MCA_PML_CALL(recv(tmpbuf[1], counts[1], datatype,
root, MCA_COLL_BASE_TAG_BCAST,
comm, MPI_STATUS_IGNORE));
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
}
/* everyone else exchanges buffers */
else {
err = ompi_coll_tuned_sendrecv( tmpbuf[lr], counts[lr], datatype,
pair, MCA_COLL_BASE_TAG_BCAST,
tmpbuf[(lr+1)%2], counts[(lr+1)%2], datatype,
pair, MCA_COLL_BASE_TAG_BCAST,
comm, MPI_STATUS_IGNORE, rank);
if (err != MPI_SUCCESS) {
line = __LINE__;
goto error_hndl;
}
}
}
return (MPI_SUCCESS);
error_hndl:
OPAL_OUTPUT((ompi_coll_tuned_stream,"%s:%4d\tError occurred %d, rank %2d", __FILE__,line,err,rank));
return (err);
}
