2011-09-11 23:02:24 +04:00
|
|
|
/*
|
Per RFC, bring in the following changes:
* Remove paffinity, maffinity, and carto frameworks -- they've been
wholly replaced by hwloc.
* Move ompi_mpi_init() affinity-setting/checking code down to ORTE.
* Update sm, smcuda, wv, and openib components to no longer use carto.
Instead, use hwloc data. There are still optimizations possible in
the sm/smcuda BTLs (i.e., making multiple mpools). Also, the old
carto-based code found out how many NUMA nodes were ''available''
-- not how many were used ''in this job''. The new hwloc-using
code computes the same value -- it was not updated to calculate how
many NUMA nodes are used ''by this job.''
* Note that I cannot compile the smcuda and wv BTLs -- I ''think''
they're right, but they need to be verified by their owners.
* The openib component now does a bunch of stuff to figure out where
"near" OpenFabrics devices are. '''THIS IS A CHANGE IN DEFAULT
BEHAVIOR!!''' and still needs to be verified by OpenFabrics vendors
(I do not have a NUMA machine with an OpenFabrics device that is a
non-uniform distance from multiple different NUMA nodes).
* Completely rewrite the OMPI_Affinity_str() routine from the
"affinity" mpiext extension. This extension now understands
hyperthreads; the output format of it has changed a bit to reflect
this new information.
* Bunches of minor changes around the code base to update names/types
from maffinity/paffinity-based names to hwloc-based names.
* Add some helper functions into the hwloc base, mainly having to do
with the fact that we have the hwloc data reporting ''all''
topology information, but sometimes you really only want the
(online | available) data.
This commit was SVN r26391.
2012-05-07 18:52:54 +04:00
|
|
|
* Copyright (c) 2011-2012 Cisco Systems, Inc. All rights reserved.
|
2011-09-11 23:02:24 +04:00
|
|
|
*
|
|
|
|
|
* $COPYRIGHT$
|
|
|
|
|
*
|
|
|
|
|
* Additional copyrights may follow
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#ifndef OPAL_MCA_HWLOC_H
|
|
|
|
|
#define OPAL_MCA_HWLOC_H
|
|
|
|
|
|
|
|
|
|
#include "opal_config.h"
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_SYS_TYPES_H
|
|
|
|
|
#include <sys/types.h>
|
|
|
|
|
#endif
|
|
|
|
|
#ifdef HAVE_SYS_TIME_H
|
|
|
|
|
#include <sys/time.h>
|
|
|
|
|
#endif
|
|
|
|
|
#ifdef HAVE_STDINT_H
|
|
|
|
|
#include <stdint.h>
|
|
|
|
|
#endif
|
|
|
|
|
#ifdef HAVE_STDARG_H
|
|
|
|
|
#include <stdarg.h>
|
|
|
|
|
#endif
|
|
|
|
|
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
#include "opal/class/opal_list.h"
|
|
|
|
|
#include "opal/class/opal_value_array.h"
|
|
|
|
|
|
2011-09-11 23:02:24 +04:00
|
|
|
#include "opal/mca/mca.h"
|
|
|
|
|
#include "opal/mca/base/base.h"
|
|
|
|
|
|
|
|
|
|
BEGIN_C_DECLS
|
|
|
|
|
|
|
|
|
|
#ifdef WIN32
|
|
|
|
|
#define WIN32_LEAN_AND_MEAN
|
|
|
|
|
#include <windows.h>
|
|
|
|
|
#undef WIN32_LEAN_AND_MEAN
|
|
|
|
|
typedef unsigned char u_char;
|
|
|
|
|
typedef unsigned short u_short;
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Structure for hwloc components.
|
|
|
|
|
*/
|
|
|
|
|
struct opal_hwloc_base_component_2_0_0_t {
|
|
|
|
|
/** MCA base component */
|
|
|
|
|
mca_base_component_t base_version;
|
|
|
|
|
/** MCA base data */
|
|
|
|
|
mca_base_component_data_t base_data;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Convenience typedef
|
|
|
|
|
*/
|
|
|
|
|
typedef struct opal_hwloc_base_component_2_0_0_t opal_hwloc_base_component_2_0_0_t;
|
|
|
|
|
typedef struct opal_hwloc_base_component_2_0_0_t opal_hwloc_component_t;
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Macro for use in components that are of type hwloc
|
|
|
|
|
*/
|
|
|
|
|
#define OPAL_HWLOC_BASE_VERSION_2_0_0 \
|
|
|
|
|
MCA_BASE_VERSION_2_0_0, \
|
|
|
|
|
"hwloc", 2, 0, 0
|
|
|
|
|
|
|
|
|
|
|
2012-09-27 03:24:27 +04:00
|
|
|
/* ******************************************************************** */
|
|
|
|
|
/* Although we cannot bind if --without-hwloc is set,
|
|
|
|
|
* we do still need to know some basic locality data
|
|
|
|
|
* like on_node and not_on_node. So ensure that we
|
|
|
|
|
* always have access to that much info by including
|
|
|
|
|
* the definitions here, outside the if-have-hwloc test
|
|
|
|
|
*/
|
|
|
|
|
typedef uint16_t opal_hwloc_locality_t;
|
|
|
|
|
#define OPAL_HWLOC_LOCALITY_T OPAL_UINT16
|
|
|
|
|
|
|
|
|
|
/** Process locality definitions */
|
|
|
|
|
enum {
|
|
|
|
|
OPAL_PROC_LOCALITY_UNKNOWN = 0x0000,
|
|
|
|
|
OPAL_PROC_NON_LOCAL = 0x8000,
|
|
|
|
|
OPAL_PROC_ON_CLUSTER = 0x0400,
|
|
|
|
|
OPAL_PROC_ON_CU = 0x0200,
|
|
|
|
|
OPAL_PROC_ON_NODE = 0x0100,
|
|
|
|
|
OPAL_PROC_ON_BOARD = 0x0080,
|
|
|
|
|
OPAL_PROC_ON_NUMA = 0x0040,
|
|
|
|
|
OPAL_PROC_ON_SOCKET = 0x0020,
|
|
|
|
|
OPAL_PROC_ON_L3CACHE = 0x0010,
|
|
|
|
|
OPAL_PROC_ON_L2CACHE = 0x0008,
|
|
|
|
|
OPAL_PROC_ON_L1CACHE = 0x0004,
|
|
|
|
|
OPAL_PROC_ON_CORE = 0x0002,
|
|
|
|
|
OPAL_PROC_ON_HWTHREAD = 0x0001,
|
|
|
|
|
OPAL_PROC_ALL_LOCAL = 0x0fff
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/** Process locality macros */
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_HWTHREAD(n) ((n) & OPAL_PROC_ON_HWTHREAD)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_CORE(n) ((n) & OPAL_PROC_ON_CORE)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_L1CACHE(n) ((n) & OPAL_PROC_ON_L1CACHE)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_L2CACHE(n) ((n) & OPAL_PROC_ON_L2CACHE)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_L3CACHE(n) ((n) & OPAL_PROC_ON_L3CACHE)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_SOCKET(n) ((n) & OPAL_PROC_ON_SOCKET)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_NUMA(n) ((n) & OPAL_PROC_ON_NUMA)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_BOARD(n) ((n) & OPAL_PROC_ON_BOARD)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_NODE(n) ((n) & OPAL_PROC_ON_NODE)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_CU(n) ((n) & OPAL_PROC_ON_CU)
|
|
|
|
|
#define OPAL_PROC_ON_LOCAL_CLUSTER(n) ((n) & OPAL_PROC_ON_CLUSTER)
|
|
|
|
|
|
|
|
|
|
/* ******************************************************************** */
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
|
Per RFC, bring in the following changes:
* Remove paffinity, maffinity, and carto frameworks -- they've been
wholly replaced by hwloc.
* Move ompi_mpi_init() affinity-setting/checking code down to ORTE.
* Update sm, smcuda, wv, and openib components to no longer use carto.
Instead, use hwloc data. There are still optimizations possible in
the sm/smcuda BTLs (i.e., making multiple mpools). Also, the old
carto-based code found out how many NUMA nodes were ''available''
-- not how many were used ''in this job''. The new hwloc-using
code computes the same value -- it was not updated to calculate how
many NUMA nodes are used ''by this job.''
* Note that I cannot compile the smcuda and wv BTLs -- I ''think''
they're right, but they need to be verified by their owners.
* The openib component now does a bunch of stuff to figure out where
"near" OpenFabrics devices are. '''THIS IS A CHANGE IN DEFAULT
BEHAVIOR!!''' and still needs to be verified by OpenFabrics vendors
(I do not have a NUMA machine with an OpenFabrics device that is a
non-uniform distance from multiple different NUMA nodes).
* Completely rewrite the OMPI_Affinity_str() routine from the
"affinity" mpiext extension. This extension now understands
hyperthreads; the output format of it has changed a bit to reflect
this new information.
* Bunches of minor changes around the code base to update names/types
from maffinity/paffinity-based names to hwloc-based names.
* Add some helper functions into the hwloc base, mainly having to do
with the fact that we have the hwloc data reporting ''all''
topology information, but sometimes you really only want the
(online | available) data.
This commit was SVN r26391.
2012-05-07 18:52:54 +04:00
|
|
|
/**
|
|
|
|
|
* Struct used to describe a section of memory (starting address
|
|
|
|
|
* and length). This is really the same thing as an iovec, but
|
|
|
|
|
* we include a separate type for it for at least 2 reasons:
|
|
|
|
|
*
|
|
|
|
|
* 1. Some OS's iovec definitions are exceedingly lame (e.g.,
|
|
|
|
|
* Solaris 9 has the length argument as an int, instead of a
|
|
|
|
|
* size_t).
|
|
|
|
|
*
|
|
|
|
|
* 2. We reserve the right to expand/change this struct in the
|
|
|
|
|
* future.
|
|
|
|
|
*/
|
|
|
|
|
typedef struct {
|
|
|
|
|
/** Starting address of segment */
|
|
|
|
|
void *mbs_start_addr;
|
|
|
|
|
/** Length of segment */
|
|
|
|
|
size_t mbs_len;
|
|
|
|
|
} opal_hwloc_base_memory_segment_t;
|
2011-10-29 18:58:58 +04:00
|
|
|
|
2012-09-27 03:24:27 +04:00
|
|
|
/* include implementation to call */
|
|
|
|
|
#if OPAL_HAVE_HWLOC
|
|
|
|
|
#include MCA_hwloc_IMPLEMENTATION_HEADER
|
|
|
|
|
|
2011-10-29 18:58:58 +04:00
|
|
|
/* define type of processor info requested */
|
|
|
|
|
typedef uint8_t opal_hwloc_resource_type_t;
|
|
|
|
|
#define OPAL_HWLOC_PHYSICAL 1
|
|
|
|
|
#define OPAL_HWLOC_LOGICAL 2
|
|
|
|
|
#define OPAL_HWLOC_AVAILABLE 3
|
|
|
|
|
|
|
|
|
|
/* structs for storing info on objects */
|
|
|
|
|
typedef struct {
|
|
|
|
|
opal_object_t super;
|
|
|
|
|
hwloc_cpuset_t available;
|
|
|
|
|
unsigned int npus;
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
unsigned int idx;
|
2012-09-15 02:01:19 +04:00
|
|
|
unsigned int num_bound;
|
2011-10-29 18:58:58 +04:00
|
|
|
} opal_hwloc_obj_data_t;
|
|
|
|
|
OBJ_CLASS_DECLARATION(opal_hwloc_obj_data_t);
|
|
|
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
|
opal_list_item_t super;
|
|
|
|
|
hwloc_obj_type_t type;
|
|
|
|
|
unsigned cache_level;
|
|
|
|
|
unsigned int num_objs;
|
|
|
|
|
opal_hwloc_resource_type_t rtype;
|
This commit introduces a new "mindist" ORTE RMAPS mapper, as well as
some relevant updates/new functionality in the opal/mca/hwloc and
orte/mca/rmaps bases. This work was mainly developed by Mellanox,
with a bunch of advice from Ralph Castain, and some minor advice from
Brice Goglin and Jeff Squyres.
Even though this is mainly Mellanox's work, Jeff is committing only
for logistical reasons (he holds the hg+svn combo tree, and can
therefore commit it directly back to SVN).
-----
Implemented distance-based mapping algorithm as a new "mindist"
component in the rmaps framework. It allows mapping processes by NUMA
due to PCI locality information as reported by the BIOS - from the
closest to device to furthest.
To use this algorithm, specify:
{{{mpirun --map-by dist:<device_name>}}}
where <device_name> can be mlx5_0, ib0, etc.
There are two modes provided:
1. bynode: load-balancing across nodes
1. byslot: go through slots sequentially (i.e., the first nodes are
more loaded)
These options are regulated by the optional ''span'' modifier; the
command line parameter looks like:
{{{mpirun --map-by dist:<device_name>,span}}}
So, for example, if there are 2 nodes, each with 8 cores, and we'd
like to run 10 processes, the mindist algorithm will place 8 processes
to the first node and 2 to the second by default. But if you want to
place 5 processes to each node, you can add a span modifier in your
command line to do that.
If there are two NUMA nodes on the node, each with 4 cores, and we run
6 processes, the mindist algorithm will try to find the NUMA closest
to the specified device, and if successful, it will place 4 processes
on that NUMA but leaving the remaining two to the next NUMA node.
You can also specify the number of cpus per MPI process. This option
is handled so that we map as many processes to the closest NUMA as we
can (number of available processors at the NUMA divided by number of
cpus per rank) and then go on with the next closest NUMA.
The default binding option for this mapping is bind-to-numa. It works
if you don't specify any binding policy. But if you specified binding
level that was "lower" than NUMA (i.e hwthread, core, socket) it would
bind to whatever level you specify.
This commit was SVN r28552.
2013-05-22 17:04:40 +04:00
|
|
|
opal_list_t sorted_by_dist_list;
|
2011-10-29 18:58:58 +04:00
|
|
|
} opal_hwloc_summary_t;
|
|
|
|
|
OBJ_CLASS_DECLARATION(opal_hwloc_summary_t);
|
|
|
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
|
opal_object_t super;
|
|
|
|
|
hwloc_cpuset_t available;
|
|
|
|
|
opal_list_t summaries;
|
2012-08-02 20:29:44 +04:00
|
|
|
|
|
|
|
|
/** \brief Additional space for custom data */
|
|
|
|
|
void *userdata;
|
2011-10-29 18:58:58 +04:00
|
|
|
} opal_hwloc_topo_data_t;
|
2012-09-04 17:52:29 +04:00
|
|
|
OPAL_DECLSPEC OBJ_CLASS_DECLARATION(opal_hwloc_topo_data_t);
|
2011-10-29 18:58:58 +04:00
|
|
|
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
/* define binding policies */
|
|
|
|
|
typedef uint16_t opal_binding_policy_t;
|
|
|
|
|
#define OPAL_BINDING_POLICY OPAL_UINT16
|
|
|
|
|
|
|
|
|
|
/* binding directives */
|
|
|
|
|
#define OPAL_BIND_IF_SUPPORTED 0x1000
|
|
|
|
|
#define OPAL_BIND_ALLOW_OVERLOAD 0x2000
|
|
|
|
|
#define OPAL_BIND_GIVEN 0x4000
|
|
|
|
|
/* binding policies */
|
|
|
|
|
#define OPAL_BIND_TO_NONE 1
|
|
|
|
|
#define OPAL_BIND_TO_BOARD 2
|
|
|
|
|
#define OPAL_BIND_TO_NUMA 3
|
|
|
|
|
#define OPAL_BIND_TO_SOCKET 4
|
|
|
|
|
#define OPAL_BIND_TO_L3CACHE 5
|
|
|
|
|
#define OPAL_BIND_TO_L2CACHE 6
|
|
|
|
|
#define OPAL_BIND_TO_L1CACHE 7
|
|
|
|
|
#define OPAL_BIND_TO_CORE 8
|
|
|
|
|
#define OPAL_BIND_TO_HWTHREAD 9
|
|
|
|
|
#define OPAL_BIND_TO_CPUSET 10
|
|
|
|
|
#define OPAL_GET_BINDING_POLICY(pol) \
|
|
|
|
|
((pol) & 0x0fff)
|
|
|
|
|
#define OPAL_SET_BINDING_POLICY(target, pol) \
|
2013-03-25 21:51:45 +04:00
|
|
|
(target) = (pol) | (((target) & 0xf000) | OPAL_BIND_GIVEN)
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
/* check if policy is set */
|
|
|
|
|
#define OPAL_BINDING_POLICY_IS_SET(pol) \
|
|
|
|
|
((pol) & 0x4000)
|
|
|
|
|
/* macro to detect if binding was qualified */
|
|
|
|
|
#define OPAL_BINDING_REQUIRED(n) \
|
|
|
|
|
(!(OPAL_BIND_IF_SUPPORTED & (n)))
|
|
|
|
|
/* macro to detect if binding is forced */
|
|
|
|
|
#define OPAL_BIND_OVERLOAD_ALLOWED(n) \
|
|
|
|
|
(OPAL_BIND_ALLOW_OVERLOAD & (n))
|
|
|
|
|
|
|
|
|
|
/* some global values */
|
2011-09-11 23:02:24 +04:00
|
|
|
OPAL_DECLSPEC extern hwloc_topology_t opal_hwloc_topology;
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
OPAL_DECLSPEC extern opal_binding_policy_t opal_hwloc_binding_policy;
|
2011-10-20 00:18:14 +04:00
|
|
|
OPAL_DECLSPEC extern hwloc_cpuset_t opal_hwloc_my_cpuset;
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
OPAL_DECLSPEC extern bool opal_hwloc_report_bindings;
|
2011-10-29 18:58:58 +04:00
|
|
|
OPAL_DECLSPEC extern hwloc_obj_type_t opal_hwloc_levels[];
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 07:40:11 +04:00
|
|
|
OPAL_DECLSPEC extern bool opal_hwloc_use_hwthreads_as_cpus;
|
2011-10-20 00:18:14 +04:00
|
|
|
|
2011-09-11 23:02:24 +04:00
|
|
|
#endif
|
|
|
|
|
|
2012-07-03 16:57:39 +04:00
|
|
|
END_C_DECLS
|
|
|
|
|
|
2011-09-11 23:02:24 +04:00
|
|
|
#endif /* OPAL_HWLOC_H_ */
|