int ompi_coll_tuned_bcast_intra_do_this(void *buf, int count,
struct ompi_datatype_t *dtype,
int root,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module,
int algorithm, int faninout, int segsize)
{
OPAL_OUTPUT((ompi_coll_tuned_stream,"coll:tuned:bcast_intra_do_this algorithm %d topo faninout %d segsize %d",
algorithm, faninout, segsize));
switch (algorithm) {
case (0):
return ompi_coll_tuned_bcast_intra_dec_fixed( buf, count, dtype, root, comm, module );
case (1):
return ompi_coll_base_bcast_intra_basic_linear( buf, count, dtype, root, comm, module );
case (2):
return ompi_coll_base_bcast_intra_chain( buf, count, dtype, root, comm, module, segsize, faninout );
case (3):
return ompi_coll_base_bcast_intra_pipeline( buf, count, dtype, root, comm, module, segsize );
case (4):
return ompi_coll_base_bcast_intra_split_bintree( buf, count, dtype, root, comm, module, segsize );
case (5):
return ompi_coll_base_bcast_intra_bintree( buf, count, dtype, root, comm, module, segsize );
case (6):
return ompi_coll_base_bcast_intra_binomial( buf, count, dtype, root, comm, module, segsize );
} /* switch */
OPAL_OUTPUT((ompi_coll_tuned_stream,"coll:tuned:bcast_intra_do_this attempt to select algorithm %d when only 0-%d is valid?",
algorithm, ompi_coll_tuned_forced_max_algorithms[BCAST]));
return (MPI_ERR_ARG);
}
/*
* bcast_intra_dec
*
* Function: - seletects broadcast algorithm to use
* Accepts: - same arguments as MPI_Bcast()
* Returns: - MPI_SUCCESS or error code (passed from the bcast implementation)
*/
int ompi_coll_tuned_bcast_intra_dec_dynamic(void *buf, int count,
struct ompi_datatype_t *dtype, int root,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
mca_coll_tuned_module_t *tuned_module = (mca_coll_tuned_module_t*) module;
OPAL_OUTPUT((ompi_coll_tuned_stream, "coll:tuned:bcast_intra_dec_dynamic"));
/* check to see if we have some filebased rules */
if (tuned_module->com_rules[BCAST]) {
/* we do, so calc the message size or what ever we need and use this for the evaluation */
int alg, faninout, segsize, ignoreme;
size_t dsize;
ompi_datatype_type_size (dtype, &dsize);
dsize *= count;
alg = ompi_coll_tuned_get_target_method_params (tuned_module->com_rules[BCAST],
dsize, &faninout, &segsize, &ignoreme);
if (alg) {
/* we have found a valid choice from the file based rules for this message size */
return ompi_coll_tuned_bcast_intra_do_this (buf, count, dtype, root,
comm, module,
alg, faninout, segsize);
} /* found a method */
} /*end if any com rules to check */
if (tuned_module->user_forced[BCAST].algorithm) {
return ompi_coll_tuned_bcast_intra_do_this(buf, count, dtype,
root, comm, module,
tuned_module->user_forced[BCAST].algorithm,
tuned_module->user_forced[BCAST].chain_fanout,
tuned_module->user_forced[BCAST].segsize);
}
return ompi_coll_tuned_bcast_intra_dec_fixed (buf, count, dtype, root,
comm, module);
}
int ompi_coll_tuned_bcast_intra_do_forced(void *buf, int count,
struct ompi_datatype_t *dtype,
int root,
struct ompi_communicator_t *comm,
mca_coll_base_module_t *module)
{
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:bcast_intra_do_forced algorithm %d",
data->user_forced[BCAST].algorithm));
switch (data->user_forced[BCAST].algorithm) {
case (0):
return ompi_coll_tuned_bcast_intra_dec_fixed( buf, count, dtype, root, comm, module );
case (1):
return ompi_coll_tuned_bcast_intra_basic_linear( buf, count, dtype, root, comm, module );
case (2):
return ompi_coll_tuned_bcast_intra_chain( buf, count, dtype, root, comm, module,
data->user_forced[BCAST].segsize,
data->user_forced[BCAST].chain_fanout );
case (3):
return ompi_coll_tuned_bcast_intra_pipeline( buf, count, dtype, root, comm, module,
data->user_forced[BCAST].segsize );
case (4):
return ompi_coll_tuned_bcast_intra_split_bintree( buf, count, dtype, root, comm, module,
data->user_forced[BCAST].segsize );
case (5):
return ompi_coll_tuned_bcast_intra_bintree( buf, count, dtype, root, comm, module,
data->user_forced[BCAST].segsize );
case (6):
return ompi_coll_tuned_bcast_intra_binomial( buf, count, dtype, root, comm, module,
data->user_forced[BCAST].segsize );
default:
OPAL_OUTPUT((ompi_coll_tuned_stream,"coll:tuned:bcast_intra_do_forced attempt to select algorithm %d when only 0-%d is valid?",
data->user_forced[BCAST].algorithm, ompi_coll_tuned_forced_max_algorithms[BCAST]));
} /* switch */
return (MPI_ERR_ARG);
}
