2015-12-18 19:10:21 +03:00
/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil -*- */
2013-07-20 02:13:58 +04:00
/*
* Copyright ( c ) 2004 - 2008 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation . All rights reserved .
* Copyright ( c ) 2004 - 2011 The University of Tennessee and The University
* of Tennessee Research Foundation . All rights
* reserved .
* Copyright ( c ) 2004 - 2005 High Performance Computing Center Stuttgart ,
* University of Stuttgart . All rights reserved .
* Copyright ( c ) 2004 - 2005 The Regents of the University of California .
* All rights reserved .
* Copyright ( c ) 2006 Sandia National Laboratories . All rights
* reserved .
2016-01-30 17:58:46 +03:00
* Copyright ( c ) 2008 - 2016 Cisco Systems , Inc . All rights reserved .
2015-12-18 19:10:21 +03:00
* Copyright ( c ) 2012 - 2016 Los Alamos National Security , LLC . All rights
2014-07-31 00:56:15 +04:00
* reserved .
2015-09-05 02:54:40 +03:00
* Copyright ( c ) 2015 Intel , Inc . All rights reserved .
2013-07-20 02:13:58 +04:00
* $ COPYRIGHT $
*
* Additional copyrights may follow
*
* $ HEADER $
*/
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
# include "opal_config.h"
2013-07-20 02:13:58 +04:00
# include <string.h>
# include <errno.h>
# include "opal/mca/base/mca_base_var.h"
# include "opal/util/argv.h"
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
# include "opal/constants.h"
2014-12-03 00:09:46 +03:00
# if BTL_IN_OPAL
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
# include "opal/mca/btl/btl.h"
# include "opal/mca/btl/base/base.h"
2014-12-03 00:09:46 +03:00
# else
# include "ompi/mca/btl/btl.h"
# include "ompi/mca/btl/base/base.h"
# endif
2013-07-20 02:13:58 +04:00
# include "btl_usnic.h"
# include "btl_usnic_frag.h"
# include "btl_usnic_endpoint.h"
# include "btl_usnic_module.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
} ;
/*
* utility routine for string parameter registration
*/
static int reg_string ( const char * param_name ,
const char * help_string ,
const char * default_value , char * * storage ,
int flags , int level )
{
* storage = ( char * ) default_value ;
mca_base_component_var_register ( & mca_btl_usnic_component . super . btl_version ,
2014-07-31 00:56:15 +04:00
param_name , help_string ,
2013-07-20 02:13:58 +04:00
MCA_BASE_VAR_TYPE_STRING ,
NULL ,
0 ,
0 ,
level ,
MCA_BASE_VAR_SCOPE_READONLY ,
storage ) ;
2014-07-31 00:56:15 +04:00
if ( 0 = = ( flags & REGSTR_EMPTY_OK ) & &
2013-07-20 02:13:58 +04:00
( NULL = = * storage | | 0 = = strlen ( * storage ) ) ) {
opal_output ( 0 , " Bad parameter value for parameter \" %s \" " ,
param_name ) ;
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
return OPAL_ERR_BAD_PARAM ;
2013-07-20 02:13:58 +04:00
}
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
return OPAL_SUCCESS ;
2013-07-20 02:13:58 +04:00
}
/*
* utility routine for integer parameter registration
*/
static int reg_int ( const char * param_name ,
const char * help_string ,
int default_value , int * storage , int flags , int level )
{
* storage = default_value ;
mca_base_component_var_register ( & mca_btl_usnic_component . super . btl_version ,
2014-07-31 00:56:15 +04:00
param_name , help_string ,
2013-07-20 02:13:58 +04:00
MCA_BASE_VAR_TYPE_INT ,
NULL ,
0 ,
0 ,
level ,
MCA_BASE_VAR_SCOPE_READONLY ,
storage ) ;
if ( 0 ! = ( flags & REGINT_NEG_ONE_OK ) & & - 1 = = * storage ) {
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
return OPAL_SUCCESS ;
2013-07-20 02:13:58 +04:00
}
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 ) ;
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
return OPAL_ERR_BAD_PARAM ;
2013-07-20 02:13:58 +04:00
}
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
return OPAL_SUCCESS ;
2013-07-20 02:13:58 +04:00
}
2014-02-27 02:21:25 +04:00
/*
* utility routine for integer parameter registration
*/
static int reg_bool ( const char * param_name ,
const char * help_string ,
bool default_value , bool * storage , int level )
{
* storage = default_value ;
mca_base_component_var_register ( & mca_btl_usnic_component . super . btl_version ,
param_name , help_string ,
MCA_BASE_VAR_TYPE_BOOL ,
NULL ,
0 ,
0 ,
level ,
MCA_BASE_VAR_SCOPE_READONLY ,
storage ) ;
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
return OPAL_SUCCESS ;
2014-02-27 02:21:25 +04:00
}
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
int opal_btl_usnic_component_register ( void )
2013-07-20 02:13:58 +04:00
{
2013-12-24 15:57:35 +04:00
int tmp , ret = 0 ;
2013-07-20 02:13:58 +04:00
static int max_modules ;
static int stats_relative ;
static int want_numa_device_assignment ;
static int sd_num ;
static int rd_num ;
static int prio_sd_num ;
static int prio_rd_num ;
static int cq_num ;
2016-01-30 17:58:46 +03:00
static int av_eq_num ;
2014-12-03 00:09:46 +03:00
static int udp_port_base ;
static int max_tiny_msg_size ;
2013-07-20 02:13:58 +04:00
static int eager_limit ;
static int rndv_eager_limit ;
2013-11-05 02:52:03 +04:00
static int pack_lazy_threshold ;
2014-08-13 19:01:20 +04:00
static int max_short_packets ;
2013-07-20 02:13:58 +04:00
# define CHECK(expr) do {\
tmp = ( expr ) ; \
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
if ( OPAL_SUCCESS ! = tmp ) ret = tmp ; \
2013-07-20 02:13:58 +04:00
} while ( 0 )
CHECK ( reg_int ( " max_btls " ,
" Maximum number of usNICs to use (default: 0 = as many as are available) " ,
0 , & max_modules ,
REGINT_GE_ZERO , OPAL_INFO_LVL_2 ) ) ;
mca_btl_usnic_component . max_modules = ( size_t ) max_modules ;
CHECK ( reg_string ( " if_include " ,
2014-12-03 00:09:46 +03:00
" Comma-delimited list of usNIC devices/networks to be used (e.g. \" eth3,usnic_0,10.10.0.0/16 \" ; empty value means to use all available usNICs). Mutually exclusive with btl_usnic_if_exclude. " ,
2014-07-31 00:56:15 +04:00
NULL , & mca_btl_usnic_component . if_include ,
2013-07-20 02:13:58 +04:00
REGSTR_EMPTY_OK , OPAL_INFO_LVL_1 ) ) ;
2014-07-31 00:56:15 +04:00
2013-07-20 02:13:58 +04:00
CHECK ( reg_string ( " if_exclude " ,
2014-12-03 00:09:46 +03:00
" Comma-delimited list of usNIC devices/networks to be excluded (empty value means to not exclude any usNICs). Mutually exclusive with btl_usnic_if_include. " ,
2013-07-20 02:13:58 +04:00
NULL , & mca_btl_usnic_component . if_exclude ,
REGSTR_EMPTY_OK , OPAL_INFO_LVL_1 ) ) ;
CHECK ( reg_int ( " stats " ,
2014-12-03 00:09:46 +03:00
" A non-negative integer specifying the frequency at which each usnic BTL will output statistics (default: 0 seconds, meaning that statistics are disabled) " ,
2013-07-20 02:13:58 +04:00
0 , & mca_btl_usnic_component . stats_frequency , 0 ,
OPAL_INFO_LVL_4 ) ) ;
2014-07-31 00:56:15 +04:00
mca_btl_usnic_component . stats_enabled =
2013-07-20 02:13:58 +04:00
( bool ) ( mca_btl_usnic_component . stats_frequency > 0 ) ;
CHECK ( reg_int ( " stats_relative " ,
2014-12-03 00:09:46 +03:00
" If stats are enabled, output relative stats between the timestamps (vs. cumulative stats since the beginning of the job) (default: 0 -- i.e., absolute) " ,
2013-07-20 02:13:58 +04:00
0 , & stats_relative , 0 , OPAL_INFO_LVL_4 ) ) ;
mca_btl_usnic_component . stats_relative = ( bool ) stats_relative ;
2015-12-18 19:10:21 +03:00
# if RCACHE_VERSION == 30
CHECK ( reg_string ( " mpool_hints " , " Hints to use when selecting mpool " ,
2016-03-17 01:04:39 +03:00
NULL , & mca_btl_usnic_component . usnic_mpool_hints ,
REGSTR_EMPTY_OK ,
2015-12-18 19:10:21 +03:00
OPAL_INFO_LVL_5 ) ) ;
CHECK ( reg_string ( " rcache " , " Name of the registration cache to be used " ,
" grdma " , & mca_btl_usnic_component . usnic_rcache_name , 0 ,
OPAL_INFO_LVL_5 ) ) ;
# else
2013-07-20 02:13:58 +04:00
CHECK ( reg_string ( " mpool " , " Name of the memory pool to be used " ,
" grdma " , & mca_btl_usnic_component . usnic_mpool_name , 0 ,
OPAL_INFO_LVL_5 ) ) ;
2015-12-18 19:10:21 +03:00
# endif
2013-07-20 02:13:58 +04:00
2015-09-18 21:23:56 +03:00
want_numa_device_assignment = OPAL_HAVE_HWLOC ? 1 : - 1 ;
2013-07-20 02:13:58 +04:00
CHECK ( reg_int ( " want_numa_device_assignment " ,
" If 1, use only Cisco VIC ports thare are a minimum NUMA distance from the MPI process for short messages. If 0, use all available Cisco VIC ports for short messages. This parameter is meaningless (and ignored) unless MPI proceses are bound to processor cores. Defaults to 1 if NUMA support is included in Open MPI; -1 otherwise. " ,
want_numa_device_assignment ,
& want_numa_device_assignment ,
0 , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . want_numa_device_assignment =
( 1 = = want_numa_device_assignment ) ? true : false ;
CHECK ( reg_int ( " sd_num " , " Maximum send descriptors to post (-1 = pre-set defaults; depends on number and type of devices available) " ,
- 1 , & sd_num , REGINT_NEG_ONE_OK , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . sd_num = ( int32_t ) sd_num ;
CHECK ( reg_int ( " rd_num " , " Number of pre-posted receive buffers (-1 = pre-set defaults; depends on number and type of devices available) " ,
- 1 , & rd_num , REGINT_NEG_ONE_OK , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . rd_num = ( int32_t ) rd_num ;
CHECK ( reg_int ( " prio_sd_num " , " Maximum priority send descriptors to post (-1 = pre-set defaults; depends on number and type of devices available) " ,
- 1 , & prio_sd_num , REGINT_NEG_ONE_OK , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . prio_sd_num = ( int32_t ) prio_sd_num ;
CHECK ( reg_int ( " prio_rd_num " , " Number of pre-posted priority receive buffers (-1 = pre-set defaults; depends on number and type of devices available) " ,
- 1 , & prio_rd_num , REGINT_NEG_ONE_OK , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . prio_rd_num = ( int32_t ) prio_rd_num ;
CHECK ( reg_int ( " cq_num " , " Number of completion queue entries (-1 = pre-set defaults; depends on number and type of devices available; will error if (sd_num+rd_num)>cq_num) " ,
- 1 , & cq_num , REGINT_NEG_ONE_OK , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . cq_num = ( int32_t ) cq_num ;
2016-01-30 21:46:14 +03:00
CHECK ( reg_int ( " av_eq_num " , " Number of event queue entries for peer address resolution " ,
2016-02-01 22:14:30 +03:00
1024 , & av_eq_num , REGINT_GE_ONE , OPAL_INFO_LVL_5 ) ) ;
2016-01-30 17:58:46 +03:00
mca_btl_usnic_component . av_eq_num = ( int32_t ) av_eq_num ;
2014-12-03 00:09:46 +03:00
CHECK ( reg_int ( " base_udp_port " , " Base UDP port to use for usNIC communications. If 0, system will pick the port number. If non-zero, it will be added to each process' local rank to obtain the final port number (default: 0) " ,
0 , & udp_port_base , REGINT_GE_ZERO , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . udp_port_base = ( int ) udp_port_base ;
2013-07-20 02:13:58 +04:00
CHECK ( reg_int ( " retrans_timeout " , " Number of microseconds before retransmitting a frame " ,
2016-01-30 21:49:14 +03:00
5000 , & mca_btl_usnic_component . retrans_timeout ,
2013-07-20 02:13:58 +04:00
REGINT_GE_ONE , OPAL_INFO_LVL_5 ) ) ;
CHECK ( reg_int ( " priority_limit " , " Max size of \" priority \" messages (0 = use pre-set defaults; depends on number and type of devices available) " ,
2014-12-03 00:09:46 +03:00
0 , & max_tiny_msg_size ,
2013-07-20 02:13:58 +04:00
REGINT_GE_ZERO , OPAL_INFO_LVL_5 ) ) ;
2014-12-03 00:09:46 +03:00
opal_btl_usnic_module_template . max_tiny_msg_size =
( size_t ) max_tiny_msg_size ;
2013-07-20 02:13:58 +04:00
CHECK ( reg_int ( " eager_limit " , " Eager send limit (0 = use pre-set defaults; depends on number and type of devices available) " ,
0 , & eager_limit , REGINT_GE_ZERO , OPAL_INFO_LVL_5 ) ) ;
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
opal_btl_usnic_module_template . super . btl_eager_limit = eager_limit ;
2013-07-20 02:13:58 +04:00
CHECK ( reg_int ( " rndv_eager_limit " , " Eager rendezvous limit (0 = use pre-set defaults; depends on number and type of devices available) " ,
0 , & rndv_eager_limit , REGINT_GE_ZERO , OPAL_INFO_LVL_5 ) ) ;
2014-07-31 00:56:15 +04:00
opal_btl_usnic_module_template . super . btl_rndv_eager_limit =
2013-07-20 02:13:58 +04:00
rndv_eager_limit ;
2013-11-05 02:52:03 +04:00
CHECK ( reg_int ( " pack_lazy_threshold " , " Convertor packing on-the-fly threshold (-1 = always pack eagerly, 0 = always pack lazily, otherwise will pack on the fly if fragment size is > limit) " ,
USNIC_DFLT_PACK_LAZY_THRESHOLD , & pack_lazy_threshold , REGINT_NEG_ONE_OK , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . pack_lazy_threshold = pack_lazy_threshold ;
2014-08-13 19:01:20 +04:00
CHECK ( reg_int ( " max_short_packets " , " Number of abnormally-short packets received before outputting a warning (0 = never show the warning) " ,
25 , & max_short_packets ,
REGINT_GE_ZERO , OPAL_INFO_LVL_5 ) ) ;
mca_btl_usnic_component . max_short_packets = max_short_packets ;
2013-07-20 02:13:58 +04:00
/* Default to bandwidth auto-detection */
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
opal_btl_usnic_module_template . super . btl_bandwidth = 0 ;
2014-12-03 00:09:46 +03:00
opal_btl_usnic_module_template . super . btl_latency = 2 ;
2013-07-20 02:13:58 +04:00
2014-06-21 00:53:50 +04:00
/* Show "cannot find route" warnings? */
mca_btl_usnic_component . show_route_failures = true ;
CHECK ( reg_bool ( " show_route_failures " ,
" Whether to show a warning when route failures between MPI process peers are detected (default = 1, enabled; 0 = disabled) " ,
mca_btl_usnic_component . show_route_failures ,
& mca_btl_usnic_component . show_route_failures ,
OPAL_INFO_LVL_3 ) ) ;
2014-02-27 02:21:25 +04:00
/* Connectivity verification */
mca_btl_usnic_component . connectivity_enabled = true ;
CHECK ( reg_bool ( " connectivity_check " ,
" Whether to enable the usNIC connectivity check upon first send (default = 1, enabled; 0 = disabled) " ,
mca_btl_usnic_component . connectivity_enabled ,
& mca_btl_usnic_component . connectivity_enabled ,
OPAL_INFO_LVL_3 ) ) ;
2014-05-02 15:06:18 +04:00
mca_btl_usnic_component . connectivity_ack_timeout = 250 ;
2014-02-27 02:21:25 +04:00
CHECK ( reg_int ( " connectivity_ack_timeout " ,
2014-12-03 00:09:46 +03:00
" Timeout, in milliseconds, while waiting for an ACK while verification connectivity between usNIC interfaces. If 0, the connectivity check is disabled (must be >=0). " ,
2014-02-27 02:21:25 +04:00
mca_btl_usnic_component . connectivity_ack_timeout ,
& mca_btl_usnic_component . connectivity_ack_timeout ,
REGINT_GE_ZERO , OPAL_INFO_LVL_3 ) ) ;
2014-05-02 15:06:18 +04:00
mca_btl_usnic_component . connectivity_num_retries = 40 ;
2014-02-27 02:21:25 +04:00
CHECK ( reg_int ( " connectivity_error_num_retries " ,
" Number of times to retry usNIC connectivity verification before aborting the MPI job (must be >0). " ,
mca_btl_usnic_component . connectivity_num_retries ,
& mca_btl_usnic_component . connectivity_num_retries ,
REGINT_GE_ONE , OPAL_INFO_LVL_3 ) ) ;
usnic: add btl_usnic_connectivity_map MCA param to output link information
If the btl_usnic_connectivity_map MCA param is set to a non-NULL
value, then each MPI process will output a file named
<prefix>-<hostname>.pid<pid>.job<jobid>.mcwrank<MCW rank>.txt. Its
contents will detail which usNIC device(s) (and therefore which
link(s)) are being used to communicate with each peer MPI process.
Here is a sample output file (named
mpi005.pid26071.job1640759297.mcwrank0.txt):
{{{
device=usnic_0,interface=eth4,ip=10.10.0.5/16,mac=24:57:20:05:20:00,mtu=9000
device=usnic_1,interface=eth5,ip=10.2.0.5/16,mac=24:57:20:05:21:00,mtu=9000
device=usnic_2,interface=eth6,ip=10.3.0.5/16,mac=24:57:20:05:50:00,mtu=9000
peer=1,hostname=mpi006,device=usnic_0@peer_ip=10.10.0.6/16@peer_mac=24:57:20:06:20:00,device=usnic_1@peer_ip=10.2.0.6/16@peer_mac=24:57:20:06:21:00,device=usnic_2@peer_ip=10.3.0.6/16@peer_mac=24:57:20:06:50:00
peer=2,hostname=mpi007,device=usnic_0@peer_ip=10.10.0.7/16@peer_mac=24:57:20:07:20:00,device=usnic_1@peer_ip=10.2.0.7/16@peer_mac=24:57:20:07:21:00,device=usnic_2@peer_ip=10.3.0.7/16@peer_mac=24:57:20:07:50:00
peer=3,hostname=mpi008,device=usnic_0@peer_ip=10.10.0.8/16@peer_mac=24:57:20:08:20:00,device=usnic_1@peer_ip=10.2.0.8/16@peer_mac=24:57:20:08:21:00,device=usnic_2@peer_ip=10.3.0.8/16@peer_mac=24:57:20:08:50:00
}}}
Reviewed by Reese Faucette
cmr=v1.8.2
This commit was SVN r32156.
2014-07-08 23:14:46 +04:00
mca_btl_usnic_component . connectivity_map_prefix = NULL ;
CHECK ( reg_string ( " connectivity_map " ,
2014-12-03 00:09:46 +03:00
" Write a per-process file containing the usNIC connectivity map. If this parameter is specified, it is the filename prefix emitted by each MPI process. The full filename emitted by each process is of the form: <prefix>-<hostname>.<pid>.<jobid>.<MCW rank>.txt. " ,
usnic: add btl_usnic_connectivity_map MCA param to output link information
If the btl_usnic_connectivity_map MCA param is set to a non-NULL
value, then each MPI process will output a file named
<prefix>-<hostname>.pid<pid>.job<jobid>.mcwrank<MCW rank>.txt. Its
contents will detail which usNIC device(s) (and therefore which
link(s)) are being used to communicate with each peer MPI process.
Here is a sample output file (named
mpi005.pid26071.job1640759297.mcwrank0.txt):
{{{
device=usnic_0,interface=eth4,ip=10.10.0.5/16,mac=24:57:20:05:20:00,mtu=9000
device=usnic_1,interface=eth5,ip=10.2.0.5/16,mac=24:57:20:05:21:00,mtu=9000
device=usnic_2,interface=eth6,ip=10.3.0.5/16,mac=24:57:20:05:50:00,mtu=9000
peer=1,hostname=mpi006,device=usnic_0@peer_ip=10.10.0.6/16@peer_mac=24:57:20:06:20:00,device=usnic_1@peer_ip=10.2.0.6/16@peer_mac=24:57:20:06:21:00,device=usnic_2@peer_ip=10.3.0.6/16@peer_mac=24:57:20:06:50:00
peer=2,hostname=mpi007,device=usnic_0@peer_ip=10.10.0.7/16@peer_mac=24:57:20:07:20:00,device=usnic_1@peer_ip=10.2.0.7/16@peer_mac=24:57:20:07:21:00,device=usnic_2@peer_ip=10.3.0.7/16@peer_mac=24:57:20:07:50:00
peer=3,hostname=mpi008,device=usnic_0@peer_ip=10.10.0.8/16@peer_mac=24:57:20:08:20:00,device=usnic_1@peer_ip=10.2.0.8/16@peer_mac=24:57:20:08:21:00,device=usnic_2@peer_ip=10.3.0.8/16@peer_mac=24:57:20:08:50:00
}}}
Reviewed by Reese Faucette
cmr=v1.8.2
This commit was SVN r32156.
2014-07-08 23:14:46 +04:00
mca_btl_usnic_component . connectivity_map_prefix ,
& mca_btl_usnic_component . connectivity_map_prefix ,
REGSTR_EMPTY_OK , OPAL_INFO_LVL_3 ) ) ;
2013-07-20 02:13:58 +04:00
return ret ;
}