#include "smpi_mvapich2_selector_stampede.h"
-static void init_mv2_alltoall_tables_stampede(){
-int i;
- int agg_table_sum = 0;
-mv2_alltoall_tuning_table **table_ptrs = NULL;
- mv2_alltoall_num_ppn_conf = 3;
- mv2_alltoall_thresholds_table
- = malloc(sizeof(mv2_alltoall_tuning_table *)
- * mv2_alltoall_num_ppn_conf);
- table_ptrs = malloc(sizeof(mv2_alltoall_tuning_table *)
- * mv2_alltoall_num_ppn_conf);
- mv2_size_alltoall_tuning_table = malloc(sizeof(int) *
- mv2_alltoall_num_ppn_conf);
- mv2_alltoall_table_ppn_conf =malloc(mv2_alltoall_num_ppn_conf * sizeof(int));
- mv2_alltoall_table_ppn_conf[0] = 1;
- mv2_size_alltoall_tuning_table[0] = 6;
- mv2_alltoall_tuning_table mv2_tmp_alltoall_thresholds_table_1ppn[] = {
- {2,
- 1,
- {{0, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {4,
- 2,
- {{0, 262144, &MPIR_Alltoall_Scatter_dest_MV2},
- {262144, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {8,
- 2,
- {{0, 8, &MPIR_Alltoall_RD_MV2},
- {8, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {16,
- 3,
- {{0, 64, &MPIR_Alltoall_RD_MV2},
- {64, 512, &MPIR_Alltoall_bruck_MV2},
- {512, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0,-1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {32,
- 3,
- {{0, 32, &MPIR_Alltoall_RD_MV2},
- {32, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {64,
- 3,
- {{0, 8, &MPIR_Alltoall_RD_MV2},
- {8, 1024, &MPIR_Alltoall_bruck_MV2},
- {1024, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
- };
- table_ptrs[0] = mv2_tmp_alltoall_thresholds_table_1ppn;
- mv2_alltoall_table_ppn_conf[1] = 2;
- mv2_size_alltoall_tuning_table[1] = 6;
- mv2_alltoall_tuning_table mv2_tmp_alltoall_thresholds_table_2ppn[] = {
- {4,
- 2,
- {{0, 32, &MPIR_Alltoall_RD_MV2},
- {32, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {8,
- 2,
- {{0, 64, &MPIR_Alltoall_RD_MV2},
- {64, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {16,
- 3,
- {{0, 64, &MPIR_Alltoall_RD_MV2},
- {64, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0,-1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {32,
- 3,
- {{0, 16, &MPIR_Alltoall_RD_MV2},
- {16, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {64,
- 3,
- {{0, 8, &MPIR_Alltoall_RD_MV2},
- {8, 1024, &MPIR_Alltoall_bruck_MV2},
- {1024, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {128,
- 3,
- {{0, 4, &MPIR_Alltoall_RD_MV2},
- {4, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{0, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
- };
- table_ptrs[1] = mv2_tmp_alltoall_thresholds_table_2ppn;
- mv2_alltoall_table_ppn_conf[2] = 16;
- mv2_size_alltoall_tuning_table[2] = 7;
- mv2_alltoall_tuning_table mv2_tmp_alltoall_thresholds_table_16ppn[] = {
- {16,
- 2,
- {{0, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{32768, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {32,
- 2,
- {{0, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_Scatter_dest_MV2},
- },
-
- {{16384, -1, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {64,
- 3,
- {{0, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, 16384, &MPIR_Alltoall_Scatter_dest_MV2},
- {16384, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{32768, 131072, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {128,
- 2,
- {{0, 2048, &MPIR_Alltoall_bruck_MV2},
- {2048, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{16384,65536, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {256,
- 2,
- {{0, 1024, &MPIR_Alltoall_bruck_MV2},
- {1024, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{16384, 65536, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- {512,
- 2,
- {{0, 1024, &MPIR_Alltoall_bruck_MV2},
- {1024, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{16384, 65536, &MPIR_Alltoall_inplace_MV2},
- },
- },
- {1024,
- 2,
- {{0, 1024, &MPIR_Alltoall_bruck_MV2},
- {1024, -1, &MPIR_Alltoall_pairwise_MV2},
- },
-
- {{16384, 65536, &MPIR_Alltoall_inplace_MV2},
- },
- },
-
- };
- table_ptrs[2] = mv2_tmp_alltoall_thresholds_table_16ppn;
- agg_table_sum = 0;
- for (i = 0; i < mv2_alltoall_num_ppn_conf; i++) {
- agg_table_sum += mv2_size_alltoall_tuning_table[i];
- }
- mv2_alltoall_thresholds_table[0] =
- malloc(agg_table_sum * sizeof (mv2_alltoall_tuning_table));
- memcpy(mv2_alltoall_thresholds_table[0], table_ptrs[0],
- (sizeof(mv2_alltoall_tuning_table)
- * mv2_size_alltoall_tuning_table[0]));
- for (i = 1; i < mv2_alltoall_num_ppn_conf; i++) {
- mv2_alltoall_thresholds_table[i] =
- mv2_alltoall_thresholds_table[i - 1]
- + mv2_size_alltoall_tuning_table[i - 1];
- memcpy(mv2_alltoall_thresholds_table[i], table_ptrs[i],
- (sizeof(mv2_alltoall_tuning_table)
- * mv2_size_alltoall_tuning_table[i]));
- }
- free(table_ptrs);
-
-
-}
-
+
int smpi_coll_tuned_alltoall_mvapich2( void *sendbuf, int sendcount,
- MPI_Datatype sendtype,
- void* recvbuf, int recvcount,
- MPI_Datatype recvtype,
- MPI_Comm comm)
+ MPI_Datatype sendtype,
+ void* recvbuf, int recvcount,
+ MPI_Datatype recvtype,
+ MPI_Comm comm)
{
- if(mv2_alltoall_table_ppn_conf==NULL)
- init_mv2_alltoall_tables_stampede();
-
- int sendtype_size, recvtype_size, nbytes, comm_size;
- char * tmp_buf = NULL;
- int mpi_errno=MPI_SUCCESS;
- int range = 0;
- int range_threshold = 0;
- int conf_index = 0;
- comm_size = smpi_comm_size(comm);
+ if(mv2_alltoall_table_ppn_conf==NULL)
+ init_mv2_alltoall_tables_stampede();
+
+ int sendtype_size, recvtype_size, nbytes, comm_size;
+ char * tmp_buf = NULL;
+ int mpi_errno=MPI_SUCCESS;
+ int range = 0;
+ int range_threshold = 0;
+ int conf_index = 0;
+ comm_size = smpi_comm_size(comm);
+
+ sendtype_size=smpi_datatype_size(sendtype);
+ recvtype_size=smpi_datatype_size(recvtype);
+ nbytes = sendtype_size * sendcount;
+
+ /* check if safe to use partial subscription mode */
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_alltoall_tuning_table[conf_index] - 1)) &&
+ (comm_size > mv2_alltoall_thresholds_table[conf_index][range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold < (mv2_alltoall_thresholds_table[conf_index][range].size_table - 1))
+ && (nbytes >
+ mv2_alltoall_thresholds_table[conf_index][range].algo_table[range_threshold].max)
+ && (mv2_alltoall_thresholds_table[conf_index][range].algo_table[range_threshold].max != -1)) {
+ range_threshold++;
+ }
+ MV2_Alltoall_function = mv2_alltoall_thresholds_table[conf_index][range].algo_table[range_threshold]
+ .MV2_pt_Alltoall_function;
+
+ if(sendbuf != MPI_IN_PLACE) {
+ mpi_errno = MV2_Alltoall_function(sendbuf, sendcount, sendtype,
+ recvbuf, recvcount, recvtype,
+ comm);
+ } else {
+ range_threshold = 0;
+ if(nbytes <
+ mv2_alltoall_thresholds_table[conf_index][range].in_place_algo_table[range_threshold].min
+ ||nbytes > mv2_alltoall_thresholds_table[conf_index][range].in_place_algo_table[range_threshold].max
+ ) {
+ tmp_buf = (char *)smpi_get_tmp_sendbuffer( comm_size * recvcount * recvtype_size );
+ mpi_errno = smpi_datatype_copy((char *)recvbuf,
+ comm_size*recvcount, recvtype,
+ (char *)tmp_buf,
+ comm_size*recvcount, recvtype);
+
+ mpi_errno = MV2_Alltoall_function(tmp_buf, recvcount, recvtype,
+ recvbuf, recvcount, recvtype,
+ comm );
+ smpi_free_tmp_buffer(tmp_buf);
+ } else {
+ mpi_errno = MPIR_Alltoall_inplace_MV2(sendbuf, sendcount, sendtype,
+ recvbuf, recvcount, recvtype,
+ comm );
+ }
+ }
+
+
+ return (mpi_errno);
+}
- sendtype_size=smpi_datatype_size(sendtype);
- recvtype_size=smpi_datatype_size(recvtype);
- nbytes = sendtype_size * sendcount;
- /* check if safe to use partial subscription mode */
- /* Search for the corresponding system size inside the tuning table */
- while ((range < (mv2_size_alltoall_tuning_table[conf_index] - 1)) &&
- (comm_size > mv2_alltoall_thresholds_table[conf_index][range].numproc)) {
- range++;
- }
- /* Search for corresponding inter-leader function */
- while ((range_threshold < (mv2_alltoall_thresholds_table[conf_index][range].size_table - 1))
- && (nbytes >
- mv2_alltoall_thresholds_table[conf_index][range].algo_table[range_threshold].max)
- && (mv2_alltoall_thresholds_table[conf_index][range].algo_table[range_threshold].max != -1)) {
- range_threshold++;
- }
- MV2_Alltoall_function = mv2_alltoall_thresholds_table[conf_index][range].algo_table[range_threshold]
- .MV2_pt_Alltoall_function;
-
- if(sendbuf != MPI_IN_PLACE) {
- mpi_errno = MV2_Alltoall_function(sendbuf, sendcount, sendtype,
- recvbuf, recvcount, recvtype,
- comm);
+int smpi_coll_tuned_allgather_mvapich2(void *sendbuf, int sendcount, MPI_Datatype sendtype,
+ void *recvbuf, int recvcount, MPI_Datatype recvtype,
+ MPI_Comm comm)
+{
+
+ int mpi_errno = MPI_SUCCESS;
+ int nbytes = 0, comm_size, recvtype_size;
+ int range = 0;
+ int partial_sub_ok = 0;
+ int conf_index = 0;
+ int range_threshold = 0;
+ int is_two_level = 0;
+ int local_size = -1;
+ MPI_Comm shmem_comm;
+ //MPI_Comm *shmem_commptr=NULL;
+ /* Get the size of the communicator */
+ comm_size = smpi_comm_size(comm);
+ recvtype_size=smpi_datatype_size(recvtype);
+ nbytes = recvtype_size * recvcount;
+
+ if(mv2_allgather_table_ppn_conf==NULL)
+ init_mv2_allgather_tables_stampede();
+
+ if(smpi_comm_get_leaders_comm(comm)==MPI_COMM_NULL){
+ smpi_comm_init_smp(comm);
+ }
+
+ int i;
+ if (smpi_comm_is_uniform(comm)){
+ shmem_comm = smpi_comm_get_intra_comm(comm);
+ local_size = smpi_comm_size(shmem_comm);
+ i = 0;
+ if (mv2_allgather_table_ppn_conf[0] == -1) {
+ // Indicating user defined tuning
+ conf_index = 0;
+ goto conf_check_end;
+ }
+ do {
+ if (local_size == mv2_allgather_table_ppn_conf[i]) {
+ conf_index = i;
+ partial_sub_ok = 1;
+ break;
+ }
+ i++;
+ } while(i < mv2_allgather_num_ppn_conf);
+ }
+ conf_check_end:
+ if (partial_sub_ok != 1) {
+ conf_index = 0;
+ }
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_allgather_tuning_table[conf_index] - 1)) &&
+ (comm_size >
+ mv2_allgather_thresholds_table[conf_index][range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold <
+ (mv2_allgather_thresholds_table[conf_index][range].size_inter_table - 1))
+ && (nbytes > mv2_allgather_thresholds_table[conf_index][range].inter_leader[range_threshold].max)
+ && (mv2_allgather_thresholds_table[conf_index][range].inter_leader[range_threshold].max !=
+ -1)) {
+ range_threshold++;
+ }
+
+ /* Set inter-leader pt */
+ MV2_Allgather_function =
+ mv2_allgather_thresholds_table[conf_index][range].inter_leader[range_threshold].
+ MV2_pt_Allgather_function;
+
+ is_two_level = mv2_allgather_thresholds_table[conf_index][range].two_level[range_threshold];
+
+ /* intracommunicator */
+ if(is_two_level ==1){
+ if(partial_sub_ok ==1){
+ if (smpi_comm_is_blocked(comm)){
+ mpi_errno = MPIR_2lvl_Allgather_MV2(sendbuf, sendcount, sendtype,
+ recvbuf, recvcount, recvtype,
+ comm);
+ }else{
+ mpi_errno = smpi_coll_tuned_allgather_mpich(sendbuf, sendcount, sendtype,
+ recvbuf, recvcount, recvtype,
+ comm);
+ }
} else {
- range_threshold = 0;
- if(nbytes <
- mv2_alltoall_thresholds_table[conf_index][range].in_place_algo_table[range_threshold].min
- ||nbytes > mv2_alltoall_thresholds_table[conf_index][range].in_place_algo_table[range_threshold].max
- ) {
- tmp_buf = (char *)malloc( comm_size * recvcount * recvtype_size );
- mpi_errno = smpi_datatype_copy((char *)recvbuf,
- comm_size*recvcount, recvtype,
- (char *)tmp_buf,
- comm_size*recvcount, recvtype);
-
- mpi_errno = MV2_Alltoall_function(tmp_buf, recvcount, recvtype,
- recvbuf, recvcount, recvtype,
- comm );
- free(tmp_buf);
- } else {
- mpi_errno = MPIR_Alltoall_inplace_MV2(sendbuf, sendcount, sendtype,
- recvbuf, recvcount, recvtype,
- comm );
- }
+ mpi_errno = MPIR_Allgather_RD_MV2(sendbuf, sendcount, sendtype,
+ recvbuf, recvcount, recvtype,
+ comm);
}
+ } else if(MV2_Allgather_function == &MPIR_Allgather_Bruck_MV2
+ || MV2_Allgather_function == &MPIR_Allgather_RD_MV2
+ || MV2_Allgather_function == &MPIR_Allgather_Ring_MV2) {
+ mpi_errno = MV2_Allgather_function(sendbuf, sendcount, sendtype,
+ recvbuf, recvcount, recvtype,
+ comm);
+ }else{
+ return MPI_ERR_OTHER;
+ }
-
- return (mpi_errno);
+ return mpi_errno;
}
-static void init_mv2_allgather_tables_stampede(){
-int i;
- int agg_table_sum = 0;
-mv2_allgather_tuning_table **table_ptrs = NULL;
- mv2_allgather_num_ppn_conf = 3;
- mv2_allgather_thresholds_table
- = malloc(sizeof(mv2_allgather_tuning_table *)
- * mv2_allgather_num_ppn_conf);
- table_ptrs = malloc(sizeof(mv2_allgather_tuning_table *)
- * mv2_allgather_num_ppn_conf);
- mv2_size_allgather_tuning_table = malloc(sizeof(int) *
- mv2_allgather_num_ppn_conf);
- mv2_allgather_table_ppn_conf
- = malloc(mv2_allgather_num_ppn_conf * sizeof(int));
- mv2_allgather_table_ppn_conf[0] = 1;
- mv2_size_allgather_tuning_table[0] = 6;
- mv2_allgather_tuning_table mv2_tmp_allgather_thresholds_table_1ppn[] = {
- {
- 2,
- {0},
- 1,
- {
- {0, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 4,
- {0,0},
- 2,
- {
- {0, 262144, &MPIR_Allgather_RD_MV2},
- {262144, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 8,
- {0,0},
- 2,
- {
- {0, 131072, &MPIR_Allgather_RD_MV2},
- {131072, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 16,
- {0,0},
- 2,
- {
- {0, 131072, &MPIR_Allgather_RD_MV2},
- {131072, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 32,
- {0,0},
- 2,
- {
- {0, 65536, &MPIR_Allgather_RD_MV2},
- {65536, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 64,
- {0,0},
- 2,
- {
- {0, 32768, &MPIR_Allgather_RD_MV2},
- {32768, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- };
- table_ptrs[0] = mv2_tmp_allgather_thresholds_table_1ppn;
- mv2_allgather_table_ppn_conf[1] = 2;
- mv2_size_allgather_tuning_table[1] = 6;
- mv2_allgather_tuning_table mv2_tmp_allgather_thresholds_table_2ppn[] = {
- {
- 4,
- {0,0},
- 2,
- {
- {0, 524288, &MPIR_Allgather_RD_MV2},
- {524288, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 8,
- {0,1,0},
- 2,
- {
- {0, 32768, &MPIR_Allgather_RD_MV2},
- {32768, 524288, &MPIR_Allgather_Ring_MV2},
- {524288, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 16,
- {0,1,0},
- 2,
- {
- {0, 16384, &MPIR_Allgather_RD_MV2},
- {16384, 524288, &MPIR_Allgather_Ring_MV2},
- {524288, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 32,
- {1,1,0},
- 2,
- {
- {0, 65536, &MPIR_Allgather_RD_MV2},
- {65536, 524288, &MPIR_Allgather_Ring_MV2},
- {524288, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 64,
- {1,1,0},
- 2,
- {
- {0, 32768, &MPIR_Allgather_RD_MV2},
- {32768, 524288, &MPIR_Allgather_Ring_MV2},
- {524288, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 128,
- {1,1,0},
- 2,
- {
- {0, 65536, &MPIR_Allgather_RD_MV2},
- {65536, 524288, &MPIR_Allgather_Ring_MV2},
- {524288, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- };
- table_ptrs[1] = mv2_tmp_allgather_thresholds_table_2ppn;
- mv2_allgather_table_ppn_conf[2] = 16;
- mv2_size_allgather_tuning_table[2] = 6;
- mv2_allgather_tuning_table mv2_tmp_allgather_thresholds_table_16ppn[] = {
- {
- 16,
- {0,0},
- 2,
- {
- {0, 1024, &MPIR_Allgather_RD_MV2},
- {1024, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 32,
- {0,0},
- 2,
- {
- {0, 1024, &MPIR_Allgather_RD_Allgather_Comm_MV2},
- {1024, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 64,
- {0,0},
- 2,
- {
- {0, 1024, &MPIR_Allgather_RD_Allgather_Comm_MV2},
- {1024, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 128,
- {0,0},
- 2,
- {
- {0, 1024, &MPIR_Allgather_RD_Allgather_Comm_MV2},
- {1024, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 256,
- {0,0},
- 2,
- {
- {0, 1024, &MPIR_Allgather_RD_Allgather_Comm_MV2},
- {1024, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
- {
- 512,
- {0,0},
- 2,
- {
- {0, 1024, &MPIR_Allgather_RD_Allgather_Comm_MV2},
- {1024, -1, &MPIR_Allgather_Ring_MV2},
- },
- },
-
- };
- table_ptrs[2] = mv2_tmp_allgather_thresholds_table_16ppn;
- agg_table_sum = 0;
- for (i = 0; i < mv2_allgather_num_ppn_conf; i++) {
- agg_table_sum += mv2_size_allgather_tuning_table[i];
- }
- mv2_allgather_thresholds_table[0] =
- malloc(agg_table_sum * sizeof (mv2_allgather_tuning_table));
- memcpy(mv2_allgather_thresholds_table[0], table_ptrs[0],
- (sizeof(mv2_allgather_tuning_table)
- * mv2_size_allgather_tuning_table[0]));
- for (i = 1; i < mv2_allgather_num_ppn_conf; i++) {
- mv2_allgather_thresholds_table[i] =
- mv2_allgather_thresholds_table[i - 1]
- + mv2_size_allgather_tuning_table[i - 1];
- memcpy(mv2_allgather_thresholds_table[i], table_ptrs[i],
- (sizeof(mv2_allgather_tuning_table)
- * mv2_size_allgather_tuning_table[i]));
- }
- free(table_ptrs);
+int smpi_coll_tuned_gather_mvapich2(void *sendbuf,
+ int sendcnt,
+ MPI_Datatype sendtype,
+ void *recvbuf,
+ int recvcnt,
+ MPI_Datatype recvtype,
+ int root, MPI_Comm comm)
+{
+ if(mv2_gather_thresholds_table==NULL)
+ init_mv2_gather_tables_stampede();
+
+ int mpi_errno = MPI_SUCCESS;
+ int range = 0;
+ int range_threshold = 0;
+ int range_intra_threshold = 0;
+ int nbytes = 0;
+ int comm_size = 0;
+ int recvtype_size, sendtype_size;
+ int rank = -1;
+ comm_size = smpi_comm_size(comm);
+ rank = smpi_comm_rank(comm);
+
+ if (rank == root) {
+ recvtype_size=smpi_datatype_size(recvtype);
+ nbytes = recvcnt * recvtype_size;
+ } else {
+ sendtype_size=smpi_datatype_size(sendtype);
+ nbytes = sendcnt * sendtype_size;
+ }
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_gather_tuning_table - 1)) &&
+ (comm_size > mv2_gather_thresholds_table[range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold < (mv2_gather_thresholds_table[range].size_inter_table - 1))
+ && (nbytes >
+ mv2_gather_thresholds_table[range].inter_leader[range_threshold].max)
+ && (mv2_gather_thresholds_table[range].inter_leader[range_threshold].max !=
+ -1)) {
+ range_threshold++;
+ }
+
+ /* Search for corresponding intra node function */
+ while ((range_intra_threshold < (mv2_gather_thresholds_table[range].size_intra_table - 1))
+ && (nbytes >
+ mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].max)
+ && (mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].max !=
+ -1)) {
+ range_intra_threshold++;
+ }
+
+ if (smpi_comm_is_blocked(comm) ) {
+ // Set intra-node function pt for gather_two_level
+ MV2_Gather_intra_node_function =
+ mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].
+ MV2_pt_Gather_function;
+ //Set inter-leader pt
+ MV2_Gather_inter_leader_function =
+ mv2_gather_thresholds_table[range].inter_leader[range_threshold].
+ MV2_pt_Gather_function;
+ // We call Gather function
+ mpi_errno =
+ MV2_Gather_inter_leader_function(sendbuf, sendcnt, sendtype, recvbuf, recvcnt,
+ recvtype, root, comm);
+
+ } else {
+ // Indeed, direct (non SMP-aware)gather is MPICH one
+ mpi_errno = smpi_coll_tuned_gather_mpich(sendbuf, sendcnt, sendtype,
+ recvbuf, recvcnt, recvtype,
+ root, comm);
+ }
+
+ return mpi_errno;
}
-int smpi_coll_tuned_allgather_mvapich2(void *sendbuf, int sendcount, MPI_Datatype sendtype,
- void *recvbuf, int recvcount, MPI_Datatype recvtype,
- MPI_Comm comm)
-{
- int mpi_errno = MPI_SUCCESS;
- int nbytes = 0, comm_size, recvtype_size;
- int range = 0;
- //int partial_sub_ok = 0;
- int conf_index = 0;
- int range_threshold = 0;
- int is_two_level = 0;
- //int local_size = -1;
- //MPI_Comm shmem_comm;
- //MPI_Comm *shmem_commptr=NULL;
- /* Get the size of the communicator */
- comm_size = smpi_comm_size(comm);
- recvtype_size=smpi_datatype_size(recvtype);
- nbytes = recvtype_size * recvcount;
-
- if(mv2_allgather_table_ppn_conf==NULL)
- init_mv2_allgather_tables_stampede();
-
- //int i;
- /* check if safe to use partial subscription mode */
- /* if (comm->ch.shmem_coll_ok == 1 && comm->ch.is_uniform) {
-
- shmem_comm = comm->ch.shmem_comm;
- MPID_Comm_get_ptr(shmem_comm, shmem_commptr);
- local_size = shmem_commptr->local_size;
- i = 0;
- if (mv2_allgather_table_ppn_conf[0] == -1) {
- // Indicating user defined tuning
- conf_index = 0;
- goto conf_check_end;
+int smpi_coll_tuned_allgatherv_mvapich2(void *sendbuf, int sendcount, MPI_Datatype sendtype,
+ void *recvbuf, int *recvcounts, int *displs,
+ MPI_Datatype recvtype, MPI_Comm comm )
+{
+ int mpi_errno = MPI_SUCCESS;
+ int range = 0, comm_size, total_count, recvtype_size, i;
+ int range_threshold = 0;
+ int nbytes = 0;
+
+ if(mv2_allgatherv_thresholds_table==NULL)
+ init_mv2_allgatherv_tables_stampede();
+
+ comm_size = smpi_comm_size(comm);
+ total_count = 0;
+ for (i = 0; i < comm_size; i++)
+ total_count += recvcounts[i];
+
+ recvtype_size=smpi_datatype_size(recvtype);
+ nbytes = total_count * recvtype_size;
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_allgatherv_tuning_table - 1)) &&
+ (comm_size > mv2_allgatherv_thresholds_table[range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold < (mv2_allgatherv_thresholds_table[range].size_inter_table - 1))
+ && (nbytes >
+ comm_size * mv2_allgatherv_thresholds_table[range].inter_leader[range_threshold].max)
+ && (mv2_allgatherv_thresholds_table[range].inter_leader[range_threshold].max !=
+ -1)) {
+ range_threshold++;
+ }
+ /* Set inter-leader pt */
+ MV2_Allgatherv_function =
+ mv2_allgatherv_thresholds_table[range].inter_leader[range_threshold].
+ MV2_pt_Allgatherv_function;
+
+ if (MV2_Allgatherv_function == &MPIR_Allgatherv_Rec_Doubling_MV2)
+ {
+ if(!(comm_size & (comm_size - 1)))
+ {
+ mpi_errno =
+ MPIR_Allgatherv_Rec_Doubling_MV2(sendbuf, sendcount,
+ sendtype, recvbuf,
+ recvcounts, displs,
+ recvtype, comm);
+ } else {
+ mpi_errno =
+ MPIR_Allgatherv_Bruck_MV2(sendbuf, sendcount,
+ sendtype, recvbuf,
+ recvcounts, displs,
+ recvtype, comm);
}
- do {
- if (local_size == mv2_allgather_table_ppn_conf[i]) {
- conf_index = i;
- partial_sub_ok = 1;
- break;
- }
- i++;
- } while(i < mv2_allgather_num_ppn_conf);
+ } else {
+ mpi_errno =
+ MV2_Allgatherv_function(sendbuf, sendcount, sendtype,
+ recvbuf, recvcounts, displs,
+ recvtype, comm);
}
- conf_check_end:
- if (partial_sub_ok != 1) {
- conf_index = 0;
- }*/
+ return mpi_errno;
+}
+
+
+
+int smpi_coll_tuned_allreduce_mvapich2(void *sendbuf,
+ void *recvbuf,
+ int count,
+ MPI_Datatype datatype,
+ MPI_Op op, MPI_Comm comm)
+{
+
+ int mpi_errno = MPI_SUCCESS;
+ //int rank = 0,
+ int comm_size = 0;
+
+ comm_size = smpi_comm_size(comm);
+ //rank = smpi_comm_rank(comm);
+
+ if (count == 0) {
+ return MPI_SUCCESS;
+ }
+
+ if (mv2_allreduce_thresholds_table == NULL)
+ init_mv2_allreduce_tables_stampede();
+
+ /* check if multiple threads are calling this collective function */
+
+ MPI_Aint sendtype_size = 0;
+ int nbytes = 0;
+ int range = 0, range_threshold = 0, range_threshold_intra = 0;
+ int is_two_level = 0;
+ int is_commutative = 0;
+ MPI_Aint true_lb, true_extent;
+
+ sendtype_size=smpi_datatype_size(datatype);
+ nbytes = count * sendtype_size;
+
+ smpi_datatype_extent(datatype, &true_lb, &true_extent);
+ //MPI_Op *op_ptr;
+ //is_commutative = smpi_op_is_commute(op);
+
+ {
/* Search for the corresponding system size inside the tuning table */
- while ((range < (mv2_size_allgather_tuning_table[conf_index] - 1)) &&
- (comm_size >
- mv2_allgather_thresholds_table[conf_index][range].numproc)) {
+ while ((range < (mv2_size_allreduce_tuning_table - 1)) &&
+ (comm_size > mv2_allreduce_thresholds_table[range].numproc)) {
range++;
}
/* Search for corresponding inter-leader function */
- while ((range_threshold <
- (mv2_allgather_thresholds_table[conf_index][range].size_inter_table - 1))
- && (nbytes > mv2_allgather_thresholds_table[conf_index][range].inter_leader[range_threshold].max)
- && (mv2_allgather_thresholds_table[conf_index][range].inter_leader[range_threshold].max !=
- -1)) {
+ /* skip mcast poiters if mcast is not available */
+ if(mv2_allreduce_thresholds_table[range].mcast_enabled != 1){
+ while ((range_threshold < (mv2_allreduce_thresholds_table[range].size_inter_table - 1))
+ && ((mv2_allreduce_thresholds_table[range].
+ inter_leader[range_threshold].MV2_pt_Allreduce_function
+ == &MPIR_Allreduce_mcst_reduce_redscat_gather_MV2) ||
+ (mv2_allreduce_thresholds_table[range].
+ inter_leader[range_threshold].MV2_pt_Allreduce_function
+ == &MPIR_Allreduce_mcst_reduce_two_level_helper_MV2)
+ )) {
+ range_threshold++;
+ }
+ }
+ while ((range_threshold < (mv2_allreduce_thresholds_table[range].size_inter_table - 1))
+ && (nbytes >
+ mv2_allreduce_thresholds_table[range].inter_leader[range_threshold].max)
+ && (mv2_allreduce_thresholds_table[range].inter_leader[range_threshold].max != -1)) {
range_threshold++;
}
+ if(mv2_allreduce_thresholds_table[range].is_two_level_allreduce[range_threshold] == 1){
+ is_two_level = 1;
+ }
+ /* Search for corresponding intra-node function */
+ while ((range_threshold_intra <
+ (mv2_allreduce_thresholds_table[range].size_intra_table - 1))
+ && (nbytes >
+ mv2_allreduce_thresholds_table[range].intra_node[range_threshold_intra].max)
+ && (mv2_allreduce_thresholds_table[range].intra_node[range_threshold_intra].max !=
+ -1)) {
+ range_threshold_intra++;
+ }
- /* Set inter-leader pt */
- MV2_Allgather_function =
- mv2_allgather_thresholds_table[conf_index][range].inter_leader[range_threshold].
- MV2_pt_Allgather_function;
-
- is_two_level = mv2_allgather_thresholds_table[conf_index][range].two_level[range_threshold];
-
- /* intracommunicator */
- if(is_two_level ==1){
-
- /* if(comm->ch.shmem_coll_ok == 1){
- MPIR_T_PVAR_COUNTER_INC(MV2, mv2_num_shmem_coll_calls, 1);
- if (1 == comm->ch.is_blocked) {
- mpi_errno = MPIR_2lvl_Allgather_MV2(sendbuf, sendcount, sendtype,
- recvbuf, recvcount, recvtype,
- comm, errflag);
- }
- else {
- mpi_errno = MPIR_Allgather_intra(sendbuf, sendcount, sendtype,
- recvbuf, recvcount, recvtype,
- comm, errflag);
- }
- } else {*/
- mpi_errno = MPIR_Allgather_RD_MV2(sendbuf, sendcount, sendtype,
- recvbuf, recvcount, recvtype,
- comm);
- // }
- } else if(MV2_Allgather_function == &MPIR_Allgather_Bruck_MV2
- || MV2_Allgather_function == &MPIR_Allgather_RD_MV2
- || MV2_Allgather_function == &MPIR_Allgather_Ring_MV2) {
- mpi_errno = MV2_Allgather_function(sendbuf, sendcount, sendtype,
- recvbuf, recvcount, recvtype,
- comm);
- }else{
- return MPI_ERR_OTHER;
+ MV2_Allreduce_function = mv2_allreduce_thresholds_table[range].inter_leader[range_threshold]
+ .MV2_pt_Allreduce_function;
+
+ MV2_Allreduce_intra_function = mv2_allreduce_thresholds_table[range].intra_node[range_threshold_intra]
+ .MV2_pt_Allreduce_function;
+
+ /* check if mcast is ready, otherwise replace mcast with other algorithm */
+ if((MV2_Allreduce_function == &MPIR_Allreduce_mcst_reduce_redscat_gather_MV2)||
+ (MV2_Allreduce_function == &MPIR_Allreduce_mcst_reduce_two_level_helper_MV2)){
+ {
+ MV2_Allreduce_function = &MPIR_Allreduce_pt2pt_rd_MV2;
+ }
+ if(is_two_level != 1) {
+ MV2_Allreduce_function = &MPIR_Allreduce_pt2pt_rd_MV2;
+ }
}
- return mpi_errno;
+ if(is_two_level == 1){
+ // check if shm is ready, if not use other algorithm first
+ if (is_commutative) {
+ if(smpi_comm_get_leaders_comm(comm)==MPI_COMM_NULL){
+ smpi_comm_init_smp(comm);
+ }
+ mpi_errno = MPIR_Allreduce_two_level_MV2(sendbuf, recvbuf, count,
+ datatype, op, comm);
+ } else {
+ mpi_errno = MPIR_Allreduce_pt2pt_rd_MV2(sendbuf, recvbuf, count,
+ datatype, op, comm);
+ }
+ } else {
+ mpi_errno = MV2_Allreduce_function(sendbuf, recvbuf, count,
+ datatype, op, comm);
+ }
+ }
+
+ //comm->ch.intra_node_done=0;
+
+ return (mpi_errno);
+
+
}
-static void init_mv2_gather_tables_stampede(){
-
- mv2_size_gather_tuning_table=7;
- mv2_gather_thresholds_table = malloc(mv2_size_gather_tuning_table*
- sizeof (mv2_gather_tuning_table));
- mv2_gather_tuning_table mv2_tmp_gather_thresholds_table[]={
- {16,
- 2,{{0, 524288, &MPIR_Gather_MV2_Direct},
- {524288, -1, &MPIR_Gather_intra}},
- 1,{{0, -1, &MPIR_Gather_MV2_Direct}}},
- {32,
- 3,{{0, 16384, &MPIR_Gather_MV2_Direct},
- {16384, 131072, &MPIR_Gather_intra},
- {131072, -1, &MPIR_Gather_MV2_two_level_Direct}},
- 1,{{0, -1, &MPIR_Gather_intra}}},
- {64,
- 3,{{0, 256, &MPIR_Gather_MV2_two_level_Direct},
- {256, 16384, &MPIR_Gather_MV2_Direct},
- {256, -1, &MPIR_Gather_MV2_two_level_Direct}},
- 1,{{0, -1, &MPIR_Gather_intra}}},
- {128,
- 3,{{0, 512, &MPIR_Gather_MV2_two_level_Direct},
- {512, 16384, &MPIR_Gather_MV2_Direct},
- {16384, -1, &MPIR_Gather_MV2_two_level_Direct}},
- 1,{{0, -1, &MPIR_Gather_intra}}},
- {256,
- 3,{{0, 512, &MPIR_Gather_MV2_two_level_Direct},
- {512, 16384, &MPIR_Gather_MV2_Direct},
- {16384, -1, &MPIR_Gather_MV2_two_level_Direct}},
- 1,{{0, -1, &MPIR_Gather_intra}}},
- {512,
- 3,{{0, 512, &MPIR_Gather_MV2_two_level_Direct},
- {512, 16384, &MPIR_Gather_MV2_Direct},
- {8196, -1, &MPIR_Gather_MV2_two_level_Direct}},
- 1,{{0, -1, &MPIR_Gather_intra}}},
- {1024,
- 3,{{0, 512, &MPIR_Gather_MV2_two_level_Direct},
- {512, 16384, &MPIR_Gather_MV2_Direct},
- {8196, -1, &MPIR_Gather_MV2_two_level_Direct}},
- 1,{{0, -1, &MPIR_Gather_intra}}},
- };
-
- memcpy(mv2_gather_thresholds_table, mv2_tmp_gather_thresholds_table,
- mv2_size_gather_tuning_table * sizeof (mv2_gather_tuning_table));
+int smpi_coll_tuned_alltoallv_mvapich2(void *sbuf, int *scounts, int *sdisps,
+ MPI_Datatype sdtype,
+ void *rbuf, int *rcounts, int *rdisps,
+ MPI_Datatype rdtype,
+ MPI_Comm comm
+)
+{
+
+ if (sbuf == MPI_IN_PLACE) {
+ return smpi_coll_tuned_alltoallv_ompi_basic_linear(sbuf, scounts, sdisps, sdtype,
+ rbuf, rcounts, rdisps,rdtype,
+ comm);
+ } else /* For starters, just keep the original algorithm. */
+ return smpi_coll_tuned_alltoallv_ring(sbuf, scounts, sdisps, sdtype,
+ rbuf, rcounts, rdisps,rdtype,
+ comm);
}
-int smpi_coll_tuned_gather_mvapich2(void *sendbuf,
- int sendcnt,
- MPI_Datatype sendtype,
- void *recvbuf,
- int recvcnt,
- MPI_Datatype recvtype,
- int root, MPI_Comm comm)
+int smpi_coll_tuned_barrier_mvapich2(MPI_Comm comm)
+{
+ return smpi_coll_tuned_barrier_mvapich2_pair(comm);
+}
+
+
+
+
+int smpi_coll_tuned_bcast_mvapich2(void *buffer,
+ int count,
+ MPI_Datatype datatype,
+ int root, MPI_Comm comm)
{
- if(mv2_alltoall_table_ppn_conf==NULL)
- init_mv2_alltoall_tables_stampede();
-
int mpi_errno = MPI_SUCCESS;
+ int comm_size/*, rank*/;
+ int two_level_bcast = 1;
+ size_t nbytes = 0;
int range = 0;
int range_threshold = 0;
- int range_intra_threshold = 0;
- int nbytes = 0;
- int comm_size = 0;
- int recvtype_size, sendtype_size;
- int rank = -1;
+ int range_threshold_intra = 0;
+ int is_homogeneous, is_contig;
+ MPI_Aint type_size;
+ //, position;
+ void *tmp_buf = NULL;
+ MPI_Comm shmem_comm;
+ //MPID_Datatype *dtp;
+
+ if (count == 0)
+ return MPI_SUCCESS;
+ if(smpi_comm_get_leaders_comm(comm)==MPI_COMM_NULL){
+ smpi_comm_init_smp(comm);
+ }
+ if(!mv2_bcast_thresholds_table)
+ init_mv2_bcast_tables_stampede();
comm_size = smpi_comm_size(comm);
- rank = smpi_comm_rank(comm);
+ //rank = smpi_comm_rank(comm);
+
+ is_contig=1;
+/* if (HANDLE_GET_KIND(datatype) == HANDLE_KIND_BUILTIN)*/
+/* is_contig = 1;*/
+/* else {*/
+/* MPID_Datatype_get_ptr(datatype, dtp);*/
+/* is_contig = dtp->is_contig;*/
+/* }*/
+
+ is_homogeneous = 1;
+
+ /* MPI_Type_size() might not give the accurate size of the packed
+ * datatype for heterogeneous systems (because of padding, encoding,
+ * etc). On the other hand, MPI_Pack_size() can become very
+ * expensive, depending on the implementation, especially for
+ * heterogeneous systems. We want to use MPI_Type_size() wherever
+ * possible, and MPI_Pack_size() in other places.
+ */
+ //if (is_homogeneous) {
+ type_size=smpi_datatype_size(datatype);
+
+ /* } else {
+ MPIR_Pack_size_impl(1, datatype, &type_size);
+ }*/
+ nbytes = (size_t) (count) * (type_size);
- if (rank == root) {
- recvtype_size=smpi_datatype_size(recvtype);
- nbytes = recvcnt * recvtype_size;
- } else {
- sendtype_size=smpi_datatype_size(sendtype);
- nbytes = sendcnt * sendtype_size;
- }
-
/* Search for the corresponding system size inside the tuning table */
- while ((range < (mv2_size_gather_tuning_table - 1)) &&
- (comm_size > mv2_gather_thresholds_table[range].numproc)) {
+ while ((range < (mv2_size_bcast_tuning_table - 1)) &&
+ (comm_size > mv2_bcast_thresholds_table[range].numproc)) {
range++;
}
/* Search for corresponding inter-leader function */
- while ((range_threshold < (mv2_gather_thresholds_table[range].size_inter_table - 1))
+ while ((range_threshold < (mv2_bcast_thresholds_table[range].size_inter_table - 1))
&& (nbytes >
- mv2_gather_thresholds_table[range].inter_leader[range_threshold].max)
- && (mv2_gather_thresholds_table[range].inter_leader[range_threshold].max !=
- -1)) {
+ mv2_bcast_thresholds_table[range].inter_leader[range_threshold].max)
+ && (mv2_bcast_thresholds_table[range].inter_leader[range_threshold].max != -1)) {
range_threshold++;
}
- /* Search for corresponding intra node function */
- while ((range_intra_threshold < (mv2_gather_thresholds_table[range].size_intra_table - 1))
+ /* Search for corresponding intra-node function */
+ while ((range_threshold_intra <
+ (mv2_bcast_thresholds_table[range].size_intra_table - 1))
&& (nbytes >
- mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].max)
- && (mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].max !=
+ mv2_bcast_thresholds_table[range].intra_node[range_threshold_intra].max)
+ && (mv2_bcast_thresholds_table[range].intra_node[range_threshold_intra].max !=
-1)) {
- range_intra_threshold++;
+ range_threshold_intra++;
+ }
+
+ MV2_Bcast_function =
+ mv2_bcast_thresholds_table[range].inter_leader[range_threshold].
+ MV2_pt_Bcast_function;
+
+ MV2_Bcast_intra_node_function =
+ mv2_bcast_thresholds_table[range].
+ intra_node[range_threshold_intra].MV2_pt_Bcast_function;
+
+/* if (mv2_user_bcast_intra == NULL && */
+/* MV2_Bcast_intra_node_function == &MPIR_Knomial_Bcast_intra_node_MV2) {*/
+/* MV2_Bcast_intra_node_function = &MPIR_Shmem_Bcast_MV2;*/
+/* }*/
+
+ if (mv2_bcast_thresholds_table[range].inter_leader[range_threshold].
+ zcpy_pipelined_knomial_factor != -1) {
+ zcpy_knomial_factor =
+ mv2_bcast_thresholds_table[range].inter_leader[range_threshold].
+ zcpy_pipelined_knomial_factor;
+ }
+
+ if (mv2_pipelined_zcpy_knomial_factor != -1) {
+ zcpy_knomial_factor = mv2_pipelined_zcpy_knomial_factor;
+ }
+
+ if(MV2_Bcast_intra_node_function == NULL) {
+ /* if tuning table do not have any intra selection, set func pointer to
+ ** default one for mcast intra node */
+ MV2_Bcast_intra_node_function = &MPIR_Shmem_Bcast_MV2;
+ }
+
+ /* Set value of pipeline segment size */
+ bcast_segment_size = mv2_bcast_thresholds_table[range].bcast_segment_size;
+
+ /* Set value of inter node knomial factor */
+ mv2_inter_node_knomial_factor = mv2_bcast_thresholds_table[range].inter_node_knomial_factor;
+
+ /* Set value of intra node knomial factor */
+ mv2_intra_node_knomial_factor = mv2_bcast_thresholds_table[range].intra_node_knomial_factor;
+
+ /* Check if we will use a two level algorithm or not */
+ two_level_bcast =
+#if defined(_MCST_SUPPORT_)
+ mv2_bcast_thresholds_table[range].is_two_level_bcast[range_threshold]
+ || comm->ch.is_mcast_ok;
+#else
+ mv2_bcast_thresholds_table[range].is_two_level_bcast[range_threshold];
+#endif
+ if (two_level_bcast == 1) {
+ if (!is_contig || !is_homogeneous) {
+ tmp_buf=(void *)smpi_get_tmp_sendbuffer(nbytes);
+
+/* position = 0;*/
+/* if (rank == root) {*/
+/* mpi_errno =*/
+/* MPIR_Pack_impl(buffer, count, datatype, tmp_buf, nbytes, &position);*/
+/* if (mpi_errno)*/
+/* MPIU_ERR_POP(mpi_errno);*/
+/* }*/
+ }
+#ifdef CHANNEL_MRAIL_GEN2
+ if ((mv2_enable_zcpy_bcast == 1) &&
+ (&MPIR_Pipelined_Bcast_Zcpy_MV2 == MV2_Bcast_function)) {
+ if (!is_contig || !is_homogeneous) {
+ mpi_errno = MPIR_Pipelined_Bcast_Zcpy_MV2(tmp_buf, nbytes, MPI_BYTE,
+ root, comm);
+ } else {
+ mpi_errno = MPIR_Pipelined_Bcast_Zcpy_MV2(buffer, count, datatype,
+ root, comm);
+ }
+ } else
+#endif /* defined(CHANNEL_MRAIL_GEN2) */
+ {
+ shmem_comm = smpi_comm_get_intra_comm(comm);
+ if (!is_contig || !is_homogeneous) {
+ mpi_errno =
+ MPIR_Bcast_tune_inter_node_helper_MV2(tmp_buf, nbytes, MPI_BYTE,
+ root, comm);
+ } else {
+ mpi_errno =
+ MPIR_Bcast_tune_inter_node_helper_MV2(buffer, count, datatype, root,
+ comm);
+ }
+
+ /* We are now done with the inter-node phase */
+
+
+ root = INTRA_NODE_ROOT;
+
+
+ if (!is_contig || !is_homogeneous) {
+ mpi_errno = MV2_Bcast_intra_node_function(tmp_buf, nbytes,
+ MPI_BYTE, root, shmem_comm);
+ } else {
+ mpi_errno = MV2_Bcast_intra_node_function(buffer, count,
+ datatype, root, shmem_comm);
+
+ }
+ }
+/* if (!is_contig || !is_homogeneous) {*/
+/* if (rank != root) {*/
+/* position = 0;*/
+/* mpi_errno = MPIR_Unpack_impl(tmp_buf, nbytes, &position, buffer,*/
+/* count, datatype);*/
+/* }*/
+/* }*/
+ } else {
+ /* We use Knomial for intra node */
+ MV2_Bcast_intra_node_function = &MPIR_Knomial_Bcast_intra_node_MV2;
+/* if (mv2_enable_shmem_bcast == 0) {*/
+ /* Fall back to non-tuned version */
+/* MPIR_Bcast_intra_MV2(buffer, count, datatype, root, comm);*/
+/* } else {*/
+ mpi_errno = MV2_Bcast_function(buffer, count, datatype, root,
+ comm);
+
+/* }*/
}
-/*
- if (comm->ch.is_global_block == 1 && mv2_use_direct_gather == 1 &&
- mv2_use_two_level_gather == 1 && comm->ch.shmem_coll_ok == 1) {
- // Set intra-node function pt for gather_two_level
- MV2_Gather_intra_node_function =
- mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].
- MV2_pt_Gather_function;
- //Set inter-leader pt
- MV2_Gather_inter_leader_function =
- mv2_gather_thresholds_table[range].inter_leader[range_threshold].
- MV2_pt_Gather_function;
- // We call Gather function
- mpi_errno =
- MV2_Gather_inter_leader_function(sendbuf, sendcnt, sendtype, recvbuf, recvcnt,
- recvtype, root, comm);
- } else {*/
- // Indded, direct (non SMP-aware)gather is MPICH one
- mpi_errno = smpi_coll_tuned_gather_mpich(sendbuf, sendcnt, sendtype,
- recvbuf, recvcnt, recvtype,
- root, comm);
- //}
return mpi_errno;
+
+}
+
+
+
+int smpi_coll_tuned_reduce_mvapich2( void *sendbuf,
+ void *recvbuf,
+ int count,
+ MPI_Datatype datatype,
+ MPI_Op op, int root, MPI_Comm comm)
+{
+ if(mv2_reduce_thresholds_table == NULL)
+ init_mv2_reduce_tables_stampede();
+
+ int mpi_errno = MPI_SUCCESS;
+ int range = 0;
+ int range_threshold = 0;
+ int range_intra_threshold = 0;
+ int is_commutative, pof2;
+ int comm_size = 0;
+ int nbytes = 0;
+ int sendtype_size;
+ int is_two_level = 0;
+
+ comm_size = smpi_comm_size(comm);
+ sendtype_size=smpi_datatype_size(datatype);
+ nbytes = count * sendtype_size;
+
+ if (count == 0)
+ return MPI_SUCCESS;
+
+ is_commutative = smpi_op_is_commute(op);
+
+ /* find nearest power-of-two less than or equal to comm_size */
+ for( pof2 = 1; pof2 <= comm_size; pof2 <<= 1 );
+ pof2 >>=1;
+
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_reduce_tuning_table - 1)) &&
+ (comm_size > mv2_reduce_thresholds_table[range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold < (mv2_reduce_thresholds_table[range].size_inter_table - 1))
+ && (nbytes >
+ mv2_reduce_thresholds_table[range].inter_leader[range_threshold].max)
+ && (mv2_reduce_thresholds_table[range].inter_leader[range_threshold].max !=
+ -1)) {
+ range_threshold++;
+ }
+
+ /* Search for corresponding intra node function */
+ while ((range_intra_threshold < (mv2_reduce_thresholds_table[range].size_intra_table - 1))
+ && (nbytes >
+ mv2_reduce_thresholds_table[range].intra_node[range_intra_threshold].max)
+ && (mv2_reduce_thresholds_table[range].intra_node[range_intra_threshold].max !=
+ -1)) {
+ range_intra_threshold++;
+ }
+
+ /* Set intra-node function pt for reduce_two_level */
+ MV2_Reduce_intra_function =
+ mv2_reduce_thresholds_table[range].intra_node[range_intra_threshold].
+ MV2_pt_Reduce_function;
+ /* Set inter-leader pt */
+ MV2_Reduce_function =
+ mv2_reduce_thresholds_table[range].inter_leader[range_threshold].
+ MV2_pt_Reduce_function;
+
+ if(mv2_reduce_intra_knomial_factor<0)
+ {
+ mv2_reduce_intra_knomial_factor = mv2_reduce_thresholds_table[range].intra_k_degree;
+ }
+ if(mv2_reduce_inter_knomial_factor<0)
+ {
+ mv2_reduce_inter_knomial_factor = mv2_reduce_thresholds_table[range].inter_k_degree;
+ }
+ if(mv2_reduce_thresholds_table[range].is_two_level_reduce[range_threshold] == 1){
+ is_two_level = 1;
+ }
+ /* We call Reduce function */
+ if(is_two_level == 1)
+ {
+ if (is_commutative == 1) {
+ if(smpi_comm_get_leaders_comm(comm)==MPI_COMM_NULL){
+ smpi_comm_init_smp(comm);
+ }
+ mpi_errno = MPIR_Reduce_two_level_helper_MV2(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ } else {
+ mpi_errno = MPIR_Reduce_binomial_MV2(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ }
+ } else if(MV2_Reduce_function == &MPIR_Reduce_inter_knomial_wrapper_MV2 ){
+ if(is_commutative ==1)
+ {
+ mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ } else {
+ mpi_errno = MPIR_Reduce_binomial_MV2(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ }
+ } else if(MV2_Reduce_function == &MPIR_Reduce_redscat_gather_MV2){
+ if (/*(HANDLE_GET_KIND(op) == HANDLE_KIND_BUILTIN) &&*/ (count >= pof2))
+ {
+ mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ } else {
+ mpi_errno = MPIR_Reduce_binomial_MV2(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ }
+ } else {
+ mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count,
+ datatype, op, root, comm);
+ }
+
+
+ return mpi_errno;
+
+}
+
+
+int smpi_coll_tuned_reduce_scatter_mvapich2(void *sendbuf, void *recvbuf, int *recvcnts,
+ MPI_Datatype datatype, MPI_Op op,
+ MPI_Comm comm)
+{
+ int mpi_errno = MPI_SUCCESS;
+ int i = 0, comm_size = smpi_comm_size(comm), total_count = 0, type_size =
+ 0, nbytes = 0;
+ int range = 0;
+ int range_threshold = 0;
+ int is_commutative = 0;
+ int *disps = xbt_malloc(comm_size * sizeof (int));
+
+ if(mv2_red_scat_thresholds_table==NULL)
+ init_mv2_reduce_scatter_tables_stampede();
+
+ is_commutative=smpi_op_is_commute(op);
+ for (i = 0; i < comm_size; i++) {
+ disps[i] = total_count;
+ total_count += recvcnts[i];
+ }
+
+ type_size=smpi_datatype_size(datatype);
+ nbytes = total_count * type_size;
+
+ if (is_commutative) {
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_red_scat_tuning_table - 1)) &&
+ (comm_size > mv2_red_scat_thresholds_table[range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold < (mv2_red_scat_thresholds_table[range].size_inter_table - 1))
+ && (nbytes >
+ mv2_red_scat_thresholds_table[range].inter_leader[range_threshold].max)
+ && (mv2_red_scat_thresholds_table[range].inter_leader[range_threshold].max !=
+ -1)) {
+ range_threshold++;
+ }
+
+ /* Set inter-leader pt */
+ MV2_Red_scat_function =
+ mv2_red_scat_thresholds_table[range].inter_leader[range_threshold].
+ MV2_pt_Red_scat_function;
+
+ mpi_errno = MV2_Red_scat_function(sendbuf, recvbuf,
+ recvcnts, datatype,
+ op, comm);
+ } else {
+ int is_block_regular = 1;
+ for (i = 0; i < (comm_size - 1); ++i) {
+ if (recvcnts[i] != recvcnts[i+1]) {
+ is_block_regular = 0;
+ break;
+ }
+ }
+ int pof2 = 1;
+ while (pof2 < comm_size) pof2 <<= 1;
+ if (pof2 == comm_size && is_block_regular) {
+ /* noncommutative, pof2 size, and block regular */
+ mpi_errno = MPIR_Reduce_scatter_non_comm_MV2(sendbuf, recvbuf,
+ recvcnts, datatype,
+ op, comm);
+ }
+ mpi_errno = smpi_coll_tuned_reduce_scatter_mpich_rdb(sendbuf, recvbuf,
+ recvcnts, datatype,
+ op, comm);
+ }
+ xbt_free(disps);
+ return mpi_errno;
+
+}
+
+
+
+int smpi_coll_tuned_scatter_mvapich2(void *sendbuf,
+ int sendcnt,
+ MPI_Datatype sendtype,
+ void *recvbuf,
+ int recvcnt,
+ MPI_Datatype recvtype,
+ int root, MPI_Comm comm)
+{
+ int range = 0, range_threshold = 0, range_threshold_intra = 0;
+ int mpi_errno = MPI_SUCCESS;
+ // int mpi_errno_ret = MPI_SUCCESS;
+ int rank, nbytes, comm_size;
+ int recvtype_size, sendtype_size;
+ int partial_sub_ok = 0;
+ int conf_index = 0;
+ int local_size = -1;
+ int i;
+ MPI_Comm shmem_comm;
+ // MPID_Comm *shmem_commptr=NULL;
+ if(mv2_scatter_thresholds_table==NULL)
+ init_mv2_scatter_tables_stampede();
+
+ if(smpi_comm_get_leaders_comm(comm)==MPI_COMM_NULL){
+ smpi_comm_init_smp(comm);
+ }
+
+ comm_size = smpi_comm_size(comm);
+
+ rank = smpi_comm_rank(comm);
+
+ if (rank == root) {
+ sendtype_size=smpi_datatype_size(sendtype);
+ nbytes = sendcnt * sendtype_size;
+ } else {
+ recvtype_size=smpi_datatype_size(recvtype);
+ nbytes = recvcnt * recvtype_size;
+ }
+
+ // check if safe to use partial subscription mode
+ if (smpi_comm_is_uniform(comm)) {
+
+ shmem_comm = smpi_comm_get_intra_comm(comm);
+ local_size = smpi_comm_size(shmem_comm);
+ i = 0;
+ if (mv2_scatter_table_ppn_conf[0] == -1) {
+ // Indicating user defined tuning
+ conf_index = 0;
+ }else{
+ do {
+ if (local_size == mv2_scatter_table_ppn_conf[i]) {
+ conf_index = i;
+ partial_sub_ok = 1;
+ break;
+ }
+ i++;
+ } while(i < mv2_scatter_num_ppn_conf);
+ }
+ }
+
+ if (partial_sub_ok != 1) {
+ conf_index = 0;
+ }
+
+ /* Search for the corresponding system size inside the tuning table */
+ while ((range < (mv2_size_scatter_tuning_table[conf_index] - 1)) &&
+ (comm_size > mv2_scatter_thresholds_table[conf_index][range].numproc)) {
+ range++;
+ }
+ /* Search for corresponding inter-leader function */
+ while ((range_threshold < (mv2_scatter_thresholds_table[conf_index][range].size_inter_table - 1))
+ && (nbytes >
+ mv2_scatter_thresholds_table[conf_index][range].inter_leader[range_threshold].max)
+ && (mv2_scatter_thresholds_table[conf_index][range].inter_leader[range_threshold].max != -1)) {
+ range_threshold++;
+ }
+
+ /* Search for corresponding intra-node function */
+ while ((range_threshold_intra <
+ (mv2_scatter_thresholds_table[conf_index][range].size_intra_table - 1))
+ && (nbytes >
+ mv2_scatter_thresholds_table[conf_index][range].intra_node[range_threshold_intra].max)
+ && (mv2_scatter_thresholds_table[conf_index][range].intra_node[range_threshold_intra].max !=
+ -1)) {
+ range_threshold_intra++;
+ }
+
+ MV2_Scatter_function = mv2_scatter_thresholds_table[conf_index][range].inter_leader[range_threshold]
+ .MV2_pt_Scatter_function;
+
+ if(MV2_Scatter_function == &MPIR_Scatter_mcst_wrap_MV2) {
+#if defined(_MCST_SUPPORT_)
+ if(comm->ch.is_mcast_ok == 1
+ && mv2_use_mcast_scatter == 1
+ && comm->ch.shmem_coll_ok == 1) {
+ MV2_Scatter_function = &MPIR_Scatter_mcst_MV2;
+ } else
+#endif /*#if defined(_MCST_SUPPORT_) */
+ {
+ if(mv2_scatter_thresholds_table[conf_index][range].inter_leader[range_threshold + 1].
+ MV2_pt_Scatter_function != NULL) {
+ MV2_Scatter_function = mv2_scatter_thresholds_table[conf_index][range].inter_leader[range_threshold + 1]
+ .MV2_pt_Scatter_function;
+ } else {
+ /* Fallback! */
+ MV2_Scatter_function = &MPIR_Scatter_MV2_Binomial;
+ }
+ }
+ }
+
+ if( (MV2_Scatter_function == &MPIR_Scatter_MV2_two_level_Direct) ||
+ (MV2_Scatter_function == &MPIR_Scatter_MV2_two_level_Binomial)) {
+ if( smpi_comm_is_blocked(comm)) {
+ MV2_Scatter_intra_function = mv2_scatter_thresholds_table[conf_index][range].intra_node[range_threshold_intra]
+ .MV2_pt_Scatter_function;
+
+ mpi_errno =
+ MV2_Scatter_function(sendbuf, sendcnt, sendtype,
+ recvbuf, recvcnt, recvtype, root,
+ comm);
+ } else {
+ mpi_errno = MPIR_Scatter_MV2_Binomial(sendbuf, sendcnt, sendtype,
+ recvbuf, recvcnt, recvtype, root,
+ comm);
+
+ }
+ } else {
+ mpi_errno = MV2_Scatter_function(sendbuf, sendcnt, sendtype,
+ recvbuf, recvcnt, recvtype, root,
+ comm);
+ }
+ return (mpi_errno);
}
