1
1
openmpi/ompi/mca/bcol/ptpcoll/bcol_ptpcoll_mca.c
Nathan Hjelm 1a021b8f2d coll/ml: add support for blocking and non-blocking allreduce, reduce, and
allgather.

The new collectives provide a signifigant performance increase over tuned for
small and medium messages. We are initially setting the priority lower than
tuned until this has had some time to soak in the trunk. Please set
coll_ml_priority to 90 for MTT runs.

Credit for this work goes to Manjunath Gorentla Venkata (ORNL), Pavel Shamis (ORNL),
and Nathan Hjelm (LANL).

Commit details (for reference):

Import ORNL's collectives for MPI_Allreduce, MPI_Reduce, and MPI_Allgather.

We need to take the basesmuma header into account when calculating the
ptpcoll small message thresholds. Add a define to bcol.h indicating the
maximum header size so we can take the header into account while not
making ptpcoll dependent on information from basesmuma.

This resolves an issue with allreduce where ptpcoll overwrites the
header of the next buffer in the basesmuma bank.

Fix reduce and make a sequential collective launcher in coll_ml_inlines.h

The root calculation for reduce was wrong for any root != 0. There are
four possibilities for the root:

 - The root is not the current process but is in the current hierarchy. In
   this case the root is the index of the global root as specified in the
   root vector.

 - The root is not the current process and is not in the next level of the
   hierarchy. In this case 0 must be the local root since this process will
   never communicate with the real root.

 - The root is not the current process but will be in next level of the
   hierarchy. In this case the current process must be the root.

 - I am the root. The root is my index.

Tested with IMB which rotates the root on every call to MPI_Reduce. Consider
IMB the reproducer for the issue this commit solves.

Make the bcast algorithm decision an enumerated variable

Resolve various asset failures when destructing coll ml requests.

Two issues:

 - Always reset the request to be invalid before returning it to the
   free list. This will avoid an asset in ompi_request_t's destructor.
   OMPI_REQUEST_FINI does this (and also releases the fortran handle
   index).

 - Never explicitly construct or destruct the superclass of an opal
   object. This screws up the class function tables and will cause
   either an assert failure or a segmentation fault when destructing
   coll ml requests.

Cleanup allgather.

I removed the duplicate non-blocking and blocking functions and modeled
the cleanup after what I found in allreduce. Also cleaned up the code
somewhat.

Don't bother copying from the send to the recieve buffer in
bcol_basesmuma_allreduce_intra_fanin_fanout if the pointers are the
same.

The eliminates a warning about memcpy and aliasing and avoids an
unnecessary call to memcpy.

Alwasy call CHECK_AND_RELEASE on memsync collectives.

There was a call to OBJ_RELEASE on the collective communicator but
because CHECK_AND_RECYLCE was never called there was not matching call
to OBJ_RELEASE. This caused coll ml to leak communicators.

Make allreduce use the sequential collective launcher in coll_ml_inlines.h

Just launch the next collective in the component progress.

I am a little unsure about this patch. There appears to be some sort
of race between collectives that causes buffer exhaustion in some cases
(IMB Allreduce is a reproducer). Changing progress to only launch the
next bcol seems to resolve the issue but might not be the best fix.

Note that I see little-no performance penalty for this change.

Fix allreduce when there are extra sources.

There was an issue with the buffer offset calculation when there are
extra sources. In the case of extra sources == 1 the offset was set
to buffer_size (just past the header of the next buffer). I adjusted
the buffer size to take into accoun the maximum header size (see the
earlier commit that added this) and simplified the offset calculation.

Make reduce/allreduce non-blocking. This is required for MPI_Comm_idup
to work correctly.

This has been tested with various layouts using the ibm testsuite and
imb and appears to have the same performance as the old blocking version.

Fix allgather for non-contiguous layouts and simplify parsing the
topology.

Some things in this patch:

 - There were several comments to the effect that level 0 of the
   hierarchy MUST contain all of the ranks. At least one function
   made this assumption but it was not true. I changed the sbgp
   components and the coll ml initization code to enforce this
   requirement.

 - Ensure that hierarchy level 0 has the ranks in the correct
   scatter gather order. This removes the need for a separate
   sort list and fixes the offset calculation for allgather.

 - There were several passes over the hierarchy to determine
   properties of the hierarchy. I eliminated these extra passes
   and the memory allocation associated with them and calculate the
   tree properties on the fly. The same DFS recursion also handles
   the re-order of level 0.

All these changes have been verified with MPI_Allreduce, MPI_Reduce, and
MPI_Allgather. All functions now pass all IBM/Open MPI, and IMB tests.

coll/ml: correct pointer usage for MPI_BOTTOM

Since contiguous datatypes are copied via memcpy (bypassing the convertor) we
need to adjust for the lb of the datatype. This corrects problems found testing
code that uses MPI_BOTTOM (NULL) as the send pointer.

Add fallback collectives for allreduce and reduce.

cmr=v1.7.5:reviewer=pasha

This commit was SVN r30363.
2014-01-22 15:39:19 +00:00

198 строки
7.0 KiB
C

/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil -*- */
/*
* Copyright (c) 2009-2012 Oak Ridge National Laboratory. All rights reserved.
* Copyright (c) 2009-2012 Mellanox Technologies. All rights reserved.
* Copyright (c) 2013 Los Alamos National Security, LLC. All rights
* reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "ompi_config.h"
#include <unistd.h>
#include <sys/types.h>
#include <sys/mman.h>
#include <fcntl.h>
#include "bcol_ptpcoll_mca.h"
#include "bcol_ptpcoll.h"
/*
* Local flags
*/
enum {
REGINT_NEG_ONE_OK = 0x01,
REGINT_GE_ZERO = 0x02,
REGINT_GE_ONE = 0x04,
REGINT_NONZERO = 0x08,
REGINT_MAX = 0x88
};
enum {
REGSTR_EMPTY_OK = 0x01,
REGSTR_MAX = 0x88
};
#if 0 /* Pasha: we will be need this function in future */
/*
* utility routine for string parameter registration
*/
static int reg_string(const char* param_name,
const char* deprecated_param_name,
const char* param_desc,
const char* default_value, char **storage,
int flags)
{
int index;
*storage = default_value;
index = mca_base_component_var_register(&mca_bcol_ptpcoll_component.super.bcol_version,
param_name, param_desc, MCA_BASE_VAR_TYPE_STRING,
NULL, 0, 0, OPAL_INFO_LVL_9,
MCA_BASE_VAR_SCOPE_READONLY, storage);
if (NULL != deprecated_param_name) {
(void) mca_base_var_register_synonym(index, "ompi", "bcol", "ptpcoll",
deprecated_param_name,
MCA_BASE_VAR_SYN_FLAG_DEPRECATED);
}
if (0 != (flags & REGSTR_EMPTY_OK) && (NULL == *storage || 0 == strlen(*storage))) {
opal_output(0, "Bad parameter value for parameter \"%s\"",
param_name);
return OMPI_ERR_BAD_PARAM;
}
return OMPI_SUCCESS;
}
#endif
/*
* utility routine for integer parameter registration
*/
static int reg_int(const char* param_name,
const char* deprecated_param_name,
const char* param_desc,
int default_value, int *storage, int flags)
{
int index;
*storage = default_value;
index = mca_base_component_var_register(&mca_bcol_ptpcoll_component.super.bcol_version,
param_name, param_desc, MCA_BASE_VAR_TYPE_INT,
NULL, 0, 0, OPAL_INFO_LVL_9,
MCA_BASE_VAR_SCOPE_READONLY, storage);
if (NULL != deprecated_param_name) {
(void) mca_base_var_register_synonym(index, "ompi", "bcol", "ptpcoll",
deprecated_param_name,
MCA_BASE_VAR_SYN_FLAG_DEPRECATED);
}
if (0 != (flags & REGINT_NEG_ONE_OK) && -1 == *storage) {
return OMPI_SUCCESS;
}
if ((0 != (flags & REGINT_GE_ZERO) && *storage < 0) ||
(0 != (flags & REGINT_GE_ONE) && *storage < 1) ||
(0 != (flags & REGINT_NONZERO) && 0 == *storage)) {
opal_output(0, "Bad parameter value for parameter \"%s\"",
param_name);
return OMPI_ERR_BAD_PARAM;
}
return OMPI_SUCCESS;
}
static int reg_bool(const char* param_name,
const char* deprecated_param_name,
const char* param_desc,
bool default_value, bool *storage)
{
int index;
*storage = default_value;
index = mca_base_component_var_register(&mca_bcol_ptpcoll_component.super.bcol_version,
param_name, param_desc, MCA_BASE_VAR_TYPE_BOOL,
NULL, 0, 0, OPAL_INFO_LVL_9,
MCA_BASE_VAR_SCOPE_READONLY, storage);
if (0 > index) {
return index;
}
if (NULL != deprecated_param_name) {
(void) mca_base_var_register_synonym(index, "ompi", "bcol", "ptpcoll",
deprecated_param_name,
MCA_BASE_VAR_SYN_FLAG_DEPRECATED);
}
return OMPI_SUCCESS;
}
int mca_bcol_ptpcoll_register_mca_params(void)
{
int ret, tmp;
mca_bcol_ptpcoll_component_t *cm = &mca_bcol_ptpcoll_component;
ret = OMPI_SUCCESS;
#define CHECK(expr) do {\
tmp = (expr); \
if (OMPI_SUCCESS != tmp) ret = tmp; \
} while (0)
CHECK(reg_int("priority", NULL,
"PTPCOLL component priority"
"(from 0(low) to 90 (high))", 90, &cm->super.priority, 0));
CHECK(reg_int("verbose", NULL,
"Output some verbose PTPCOLL information "
"(0 = no output, nonzero = output)", 0, &cm->verbose, REGINT_GE_ZERO));
CHECK(reg_int("k_nomial_radix", NULL,
"The radix of K-Nomial Tree "
"(starts from 2)", 2, &cm->k_nomial_radix, REGINT_GE_ONE));
CHECK(reg_int("narray_radix", NULL,
"The radix of Narray Tree "
"(starts from 2)", 2, &cm->narray_radix, REGINT_GE_ONE));
CHECK(reg_int("narray_knomial_radix", NULL,
"The radix of Narray/Knomial Tree for scatther-gather type algorithms"
"(starts from 2)", 2, &cm->narray_knomial_radix, REGINT_GE_ONE));
CHECK(reg_int("num_to_probe", NULL,
"Number of probe operation in single source data check"
"(starts from 8)", 8, &cm->num_to_probe, REGINT_GE_ONE));
CHECK(reg_int("bcast_small_msg_known_root_alg", NULL,
"Algoritm selection for bcast small messages known root"
"(1 - K-nomial, 2 - N-array)", 1, &cm->bcast_small_messages_known_root_alg,
REGINT_GE_ZERO));
CHECK(reg_int("bcast_large_msg_known_root_alg", NULL,
"Algoritm selection for bcast large messages known root"
"(1 - Binomial scatther-gather, 2 - N-array scather, K-nomial gather)",
1, &cm->bcast_large_messages_known_root_alg, REGINT_GE_ZERO));
CHECK(reg_int("barrier_alg", NULL,
"Algoritm selection for Barrier"
"(1 - Recursive doubling, 2 - Recursive K-ing)",
1, &cm->barrier_alg, REGINT_GE_ZERO));
/* register parmeters controlling message fragementation */
CHECK(reg_int("min_frag_size", NULL,
"Minimum fragment size",
getpagesize(), &cm->super.min_frag_size, REGINT_GE_ONE));
CHECK(reg_int("max_frag_size", NULL,
"Maximum fragment size",
FRAG_SIZE_NO_LIMIT, &cm->super.max_frag_size, REGINT_NONZERO));
CHECK(reg_bool("can_use_user_buffers", NULL,
"User memory can be used by the collective algorithms",
1, &cm->super.can_use_user_buffers));
return ret;
}