2008-10-28 20:22:29 +03:00
# -*- shell-script -*-
#
# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
# University Research and Technology
# Corporation. All rights reserved.
# Copyright (c) 2004-2005 The University of Tennessee and The University
# of Tennessee Research Foundation. All rights
# reserved.
2015-06-24 06:59:57 +03:00
# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
2008-10-28 20:22:29 +03:00
# University of Stuttgart. All rights reserved.
# Copyright (c) 2004-2005 The Regents of the University of California.
# All rights reserved.
2015-01-17 05:13:31 +03:00
# Copyright (c) 2006-2015 Cisco Systems, Inc. All rights reserved.
2015-01-06 00:36:57 +03:00
# Copyright (c) 2006-2015 Los Alamos National Security, LLC. All rights
2008-10-28 20:22:29 +03:00
# reserved.
2009-12-15 16:52:19 +03:00
# Copyright (c) 2006-2009 Mellanox Technologies. All rights reserved.
2012-04-03 15:02:34 +04:00
# Copyright (c) 2010-2012 Oracle and/or its affiliates. All rights reserved.
2012-07-02 19:20:12 +04:00
# Copyright (c) 2009-2012 Oak Ridge National Laboratory. All rights reserved.
2014-12-09 12:43:15 +03:00
# Copyright (c) 2014 Bull SAS. All rights reserved.
# Copyright (c) 2014-2015 Research Organization for Information Science
# and Technology (RIST). All rights reserved.
2008-10-28 20:22:29 +03:00
# $COPYRIGHT$
2015-06-24 06:59:57 +03:00
#
2008-10-28 20:22:29 +03:00
# Additional copyrights may follow
2015-06-24 06:59:57 +03:00
#
2008-10-28 20:22:29 +03:00
# $HEADER$
#
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
# OPAL_CHECK_OPENFABRICS(prefix, [action-if-found], [action-if-not-found])
2008-10-28 20:22:29 +03:00
# --------------------------------------------------------
2015-06-24 06:59:57 +03:00
# check if OPENIB support can be found. sets prefix_{CPPFLAGS,
2008-10-28 20:22:29 +03:00
# LDFLAGS, LIBS} as needed and runs action-if-found if there is
# support, otherwise executes action-if-not-found
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
AC_DEFUN([OPAL_CHECK_OPENFABRICS],[
2010-11-13 02:22:11 +03:00
OPAL_VAR_SCOPE_PUSH([$1_msg])
2008-10-28 20:22:29 +03:00
This commit unifies the configury of some verbs-lovin' components.
* Add new configure command line options and deprecate some old ones:
* --with-verbs replaces --with-openib
* --with-verbs-libdir replaces --with-openib-libdir
* If you specify --with-openib[-libdir] without
--with-verbs[-libdir], you'll get a "these options have been
deprecated!" warning, but then they'll act just like
--with-verbs[--libdir].
'''Sidenote:''' Note that we are not renaming any components at this
time, nor are we renaming the top-level OMPI_CHECK_OPENIB m4 macro
(which is pretty strongly tied to the openib BTL and is bastaridzed
by the ofud BTL). Note that there will likely be more changes in
this area coming soon (next week?) when some long-standing changes
move to the SVN trunk: some openib BTL infrastructure will move to
ompi/mca/common, and its configury gets split up / refactored.
We extend our philosophy of other --with-<foo> configure options of
--with-verbs to ''all'' verbs-lovin components:
* If you specify --with-verbs, then all verbs-lovin' components must
configure successfully (or abort). This currently means: OOB ud,
BTL ofud, BTL openib.
* If you specify --with-verbs=DIR, then all verbs-lovin' component
must configure successfully (or abort), and will use DIR to find
verbs headers and libraries.
* If you specify --without-verbs, then all verbs-lovin' components
will be ignored.
This commit also fixes a problem where the --with-openib=DIR form
would not use DIR for ''all'' verbs-lovin' components (I think only
BTL openib and BTL ofud used that DIR). Now all of them do, as does
hwloc (because hwloc has some !OpenFabrics helper functions that
require ibv types from verbs.h).
There's a little new m4 infrastructure worth mentioning:
* If you create a new verbs-lovin' component (i.e., a component that
need verbs), your configure.m4 should
AC_REQUIRE([OPAL_CHECK_VERBS_DIR]).
* You can then use three global shell variables: $opal_want_verbs,
$opal_verbs_dir, $opal_verbs_libdir, which will be set as follows:
* opal_want_verbs will be "yes" and opal_verbs_dir and
opal_verbs_libdir will both be set to directory values, '''OR'''
* opal_want_verbs will be "no" and opal_verbs_dir and
opal_verbs_libdir will both be set empty
This commit was SVN r26640.
2012-06-22 23:53:56 +04:00
# Setup the --with switches to allow users to specify where
# verbs stuff lives.
AC_REQUIRE([OPAL_CHECK_VERBS_DIR])
2008-10-28 20:22:29 +03:00
2009-05-27 07:03:18 +04:00
#
# Add padding to OpenIB header
#
AC_ARG_ENABLE([openib-control-hdr-padding],
[AC_HELP_STRING([--enable-openib-control-hdr-padding],
This commit unifies the configury of some verbs-lovin' components.
* Add new configure command line options and deprecate some old ones:
* --with-verbs replaces --with-openib
* --with-verbs-libdir replaces --with-openib-libdir
* If you specify --with-openib[-libdir] without
--with-verbs[-libdir], you'll get a "these options have been
deprecated!" warning, but then they'll act just like
--with-verbs[--libdir].
'''Sidenote:''' Note that we are not renaming any components at this
time, nor are we renaming the top-level OMPI_CHECK_OPENIB m4 macro
(which is pretty strongly tied to the openib BTL and is bastaridzed
by the ofud BTL). Note that there will likely be more changes in
this area coming soon (next week?) when some long-standing changes
move to the SVN trunk: some openib BTL infrastructure will move to
ompi/mca/common, and its configury gets split up / refactored.
We extend our philosophy of other --with-<foo> configure options of
--with-verbs to ''all'' verbs-lovin components:
* If you specify --with-verbs, then all verbs-lovin' components must
configure successfully (or abort). This currently means: OOB ud,
BTL ofud, BTL openib.
* If you specify --with-verbs=DIR, then all verbs-lovin' component
must configure successfully (or abort), and will use DIR to find
verbs headers and libraries.
* If you specify --without-verbs, then all verbs-lovin' components
will be ignored.
This commit also fixes a problem where the --with-openib=DIR form
would not use DIR for ''all'' verbs-lovin' components (I think only
BTL openib and BTL ofud used that DIR). Now all of them do, as does
hwloc (because hwloc has some !OpenFabrics helper functions that
require ibv types from verbs.h).
There's a little new m4 infrastructure worth mentioning:
* If you create a new verbs-lovin' component (i.e., a component that
need verbs), your configure.m4 should
AC_REQUIRE([OPAL_CHECK_VERBS_DIR]).
* You can then use three global shell variables: $opal_want_verbs,
$opal_verbs_dir, $opal_verbs_libdir, which will be set as follows:
* opal_want_verbs will be "yes" and opal_verbs_dir and
opal_verbs_libdir will both be set to directory values, '''OR'''
* opal_want_verbs will be "no" and opal_verbs_dir and
opal_verbs_libdir will both be set empty
This commit was SVN r26640.
2012-06-22 23:53:56 +04:00
[Add padding bytes to the openib BTL control header (default:disabled)])])
AC_MSG_CHECKING([if want to add padding to the openib BTL control header])
2009-05-27 07:03:18 +04:00
if test "$enable_openib_control_hdr_padding" = "yes"; then
AC_MSG_RESULT([yes])
ompi_openib_pad_hdr=1
2012-04-03 15:02:34 +04:00
elif test "$enable_openib_control_hdr_padding" = "no"; then
2009-05-27 07:03:18 +04:00
AC_MSG_RESULT([no])
ompi_openib_pad_hdr=0
2012-04-03 15:02:34 +04:00
else
#
# Enable padding for SPARC platforms by default because the
2015-06-24 06:59:57 +03:00
# btl will segv otherwise. Keep padding disabled for other
2012-04-03 15:02:34 +04:00
# platforms since there are some performance implications with
# padding on for those plaforms.
#
case "${host}" in
sparc*)
AC_MSG_RESULT([yes (enabled by default on SPARC)])
ompi_openib_pad_hdr=1
;;
*)
AC_MSG_RESULT([no])
ompi_openib_pad_hdr=0
;;
esac
2009-05-27 07:03:18 +04:00
fi
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
AC_DEFINE_UNQUOTED([OPAL_OPENIB_PAD_HDR], [$ompi_openib_pad_hdr],
This commit unifies the configury of some verbs-lovin' components.
* Add new configure command line options and deprecate some old ones:
* --with-verbs replaces --with-openib
* --with-verbs-libdir replaces --with-openib-libdir
* If you specify --with-openib[-libdir] without
--with-verbs[-libdir], you'll get a "these options have been
deprecated!" warning, but then they'll act just like
--with-verbs[--libdir].
'''Sidenote:''' Note that we are not renaming any components at this
time, nor are we renaming the top-level OMPI_CHECK_OPENIB m4 macro
(which is pretty strongly tied to the openib BTL and is bastaridzed
by the ofud BTL). Note that there will likely be more changes in
this area coming soon (next week?) when some long-standing changes
move to the SVN trunk: some openib BTL infrastructure will move to
ompi/mca/common, and its configury gets split up / refactored.
We extend our philosophy of other --with-<foo> configure options of
--with-verbs to ''all'' verbs-lovin components:
* If you specify --with-verbs, then all verbs-lovin' components must
configure successfully (or abort). This currently means: OOB ud,
BTL ofud, BTL openib.
* If you specify --with-verbs=DIR, then all verbs-lovin' component
must configure successfully (or abort), and will use DIR to find
verbs headers and libraries.
* If you specify --without-verbs, then all verbs-lovin' components
will be ignored.
This commit also fixes a problem where the --with-openib=DIR form
would not use DIR for ''all'' verbs-lovin' components (I think only
BTL openib and BTL ofud used that DIR). Now all of them do, as does
hwloc (because hwloc has some !OpenFabrics helper functions that
require ibv types from verbs.h).
There's a little new m4 infrastructure worth mentioning:
* If you create a new verbs-lovin' component (i.e., a component that
need verbs), your configure.m4 should
AC_REQUIRE([OPAL_CHECK_VERBS_DIR]).
* You can then use three global shell variables: $opal_want_verbs,
$opal_verbs_dir, $opal_verbs_libdir, which will be set as follows:
* opal_want_verbs will be "yes" and opal_verbs_dir and
opal_verbs_libdir will both be set to directory values, '''OR'''
* opal_want_verbs will be "no" and opal_verbs_dir and
opal_verbs_libdir will both be set empty
This commit was SVN r26640.
2012-06-22 23:53:56 +04:00
[Add padding bytes to the openib BTL control header])
2009-05-27 07:03:18 +04:00
This commit unifies the configury of some verbs-lovin' components.
* Add new configure command line options and deprecate some old ones:
* --with-verbs replaces --with-openib
* --with-verbs-libdir replaces --with-openib-libdir
* If you specify --with-openib[-libdir] without
--with-verbs[-libdir], you'll get a "these options have been
deprecated!" warning, but then they'll act just like
--with-verbs[--libdir].
'''Sidenote:''' Note that we are not renaming any components at this
time, nor are we renaming the top-level OMPI_CHECK_OPENIB m4 macro
(which is pretty strongly tied to the openib BTL and is bastaridzed
by the ofud BTL). Note that there will likely be more changes in
this area coming soon (next week?) when some long-standing changes
move to the SVN trunk: some openib BTL infrastructure will move to
ompi/mca/common, and its configury gets split up / refactored.
We extend our philosophy of other --with-<foo> configure options of
--with-verbs to ''all'' verbs-lovin components:
* If you specify --with-verbs, then all verbs-lovin' components must
configure successfully (or abort). This currently means: OOB ud,
BTL ofud, BTL openib.
* If you specify --with-verbs=DIR, then all verbs-lovin' component
must configure successfully (or abort), and will use DIR to find
verbs headers and libraries.
* If you specify --without-verbs, then all verbs-lovin' components
will be ignored.
This commit also fixes a problem where the --with-openib=DIR form
would not use DIR for ''all'' verbs-lovin' components (I think only
BTL openib and BTL ofud used that DIR). Now all of them do, as does
hwloc (because hwloc has some !OpenFabrics helper functions that
require ibv types from verbs.h).
There's a little new m4 infrastructure worth mentioning:
* If you create a new verbs-lovin' component (i.e., a component that
need verbs), your configure.m4 should
AC_REQUIRE([OPAL_CHECK_VERBS_DIR]).
* You can then use three global shell variables: $opal_want_verbs,
$opal_verbs_dir, $opal_verbs_libdir, which will be set as follows:
* opal_want_verbs will be "yes" and opal_verbs_dir and
opal_verbs_libdir will both be set to directory values, '''OR'''
* opal_want_verbs will be "no" and opal_verbs_dir and
opal_verbs_libdir will both be set empty
This commit was SVN r26640.
2012-06-22 23:53:56 +04:00
AS_IF([test "$opal_want_verbs" = "no"],
2015-10-13 08:12:52 +03:00
[opal_check_openib_happy="no"],
[opal_check_openib_happy="yes"])
2008-10-28 20:22:29 +03:00
ompi_check_openib_$1_save_CPPFLAGS="$CPPFLAGS"
ompi_check_openib_$1_save_LDFLAGS="$LDFLAGS"
ompi_check_openib_$1_save_LIBS="$LIBS"
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2008-10-28 20:22:29 +03:00
[AC_CHECK_HEADERS(
fcntl.h sys/poll.h,
[],
[AC_MSG_WARN([fcntl.h sys/poll.h not found. Can not build component.])
2015-10-13 08:12:52 +03:00
opal_check_openib_happy="no"])])
2008-10-28 20:22:29 +03:00
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2014-05-01 18:24:56 +04:00
[OPAL_CHECK_PACKAGE([$1],
2008-10-28 20:22:29 +03:00
[infiniband/verbs.h],
[ibverbs],
[ibv_open_device],
[],
This commit unifies the configury of some verbs-lovin' components.
* Add new configure command line options and deprecate some old ones:
* --with-verbs replaces --with-openib
* --with-verbs-libdir replaces --with-openib-libdir
* If you specify --with-openib[-libdir] without
--with-verbs[-libdir], you'll get a "these options have been
deprecated!" warning, but then they'll act just like
--with-verbs[--libdir].
'''Sidenote:''' Note that we are not renaming any components at this
time, nor are we renaming the top-level OMPI_CHECK_OPENIB m4 macro
(which is pretty strongly tied to the openib BTL and is bastaridzed
by the ofud BTL). Note that there will likely be more changes in
this area coming soon (next week?) when some long-standing changes
move to the SVN trunk: some openib BTL infrastructure will move to
ompi/mca/common, and its configury gets split up / refactored.
We extend our philosophy of other --with-<foo> configure options of
--with-verbs to ''all'' verbs-lovin components:
* If you specify --with-verbs, then all verbs-lovin' components must
configure successfully (or abort). This currently means: OOB ud,
BTL ofud, BTL openib.
* If you specify --with-verbs=DIR, then all verbs-lovin' component
must configure successfully (or abort), and will use DIR to find
verbs headers and libraries.
* If you specify --without-verbs, then all verbs-lovin' components
will be ignored.
This commit also fixes a problem where the --with-openib=DIR form
would not use DIR for ''all'' verbs-lovin' components (I think only
BTL openib and BTL ofud used that DIR). Now all of them do, as does
hwloc (because hwloc has some !OpenFabrics helper functions that
require ibv types from verbs.h).
There's a little new m4 infrastructure worth mentioning:
* If you create a new verbs-lovin' component (i.e., a component that
need verbs), your configure.m4 should
AC_REQUIRE([OPAL_CHECK_VERBS_DIR]).
* You can then use three global shell variables: $opal_want_verbs,
$opal_verbs_dir, $opal_verbs_libdir, which will be set as follows:
* opal_want_verbs will be "yes" and opal_verbs_dir and
opal_verbs_libdir will both be set to directory values, '''OR'''
* opal_want_verbs will be "no" and opal_verbs_dir and
opal_verbs_libdir will both be set empty
This commit was SVN r26640.
2012-06-22 23:53:56 +04:00
[$opal_verbs_dir],
[$opal_verbs_libdir],
2015-10-13 08:12:52 +03:00
[opal_check_openib_happy="yes"],
[opal_check_openib_happy="no"])])
2008-10-28 20:22:29 +03:00
CPPFLAGS="$CPPFLAGS $$1_CPPFLAGS"
LDFLAGS="$LDFLAGS $$1_LDFLAGS"
LIBS="$LIBS $$1_LIBS"
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2008-10-28 20:22:29 +03:00
[AC_CACHE_CHECK(
[number of arguments to ibv_create_cq],
[ompi_cv_func_ibv_create_cq_args],
[AC_LINK_IFELSE(
[AC_LANG_PROGRAM(
[[#include <infiniband/verbs.h> ]],
[[ibv_create_cq(NULL, 0, NULL, NULL, 0);]])],
[ompi_cv_func_ibv_create_cq_args=5],
[AC_LINK_IFELSE(
[AC_LANG_PROGRAM(
[[#include <infiniband/verbs.h> ]],
[[ibv_create_cq(NULL, 0, NULL);]])],
[ompi_cv_func_ibv_create_cq_args=3],
[ompi_cv_func_ibv_create_cq_args="unknown"])])])
AS_IF([test "$ompi_cv_func_ibv_create_cq_args" = "unknown"],
[AC_MSG_WARN([Can not determine number of args to ibv_create_cq.])
AC_MSG_WARN([Not building component.])
2015-10-13 08:12:52 +03:00
opal_check_openib_happy="no"],
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
[AC_DEFINE_UNQUOTED([OPAL_IBV_CREATE_CQ_ARGS],
2008-10-28 20:22:29 +03:00
[$ompi_cv_func_ibv_create_cq_args],
[Number of arguments to ibv_create_cq])])])
2011-06-28 18:28:29 +04:00
#
# OpenIB dynamic SL
#
AC_ARG_ENABLE([openib-dynamic-sl],
[AC_HELP_STRING([--enable-openib-dynamic-sl],
2011-07-14 03:19:58 +04:00
[Enable openib BTL to query Subnet Manager for IB SL (default: enabled)])])
2011-06-28 18:28:29 +04:00
2008-10-28 20:22:29 +03:00
# Set these up so that we can do an AC_DEFINE below
# (unconditionally)
$1_have_xrc=0
2015-01-07 07:27:25 +03:00
$1_have_xrc_domains=0
2011-07-03 18:02:50 +04:00
$1_have_opensm_devel=0
2008-10-28 20:22:29 +03:00
# If we have the openib stuff available, find out what we've got
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2015-01-06 00:36:57 +03:00
[AC_CHECK_DECLS([IBV_EVENT_CLIENT_REREGISTER, IBV_ACCESS_SO, IBV_ATOMIC_HCA], [], [],
2008-10-28 20:22:29 +03:00
[#include <infiniband/verbs.h>])
AC_CHECK_FUNCS([ibv_get_device_list ibv_resize_cq])
# struct ibv_device.transport_type was added in OFED v1.2
AC_CHECK_MEMBERS([struct ibv_device.transport_type], [], [],
[#include <infiniband/verbs.h>])
2015-07-13 03:56:51 +03:00
# We have to check functions both exits *and* are declared
# since some distro ship broken ibverbs devel headers
# IBV_DEVICE_XRC is common to all OFED versions
2008-10-28 20:22:29 +03:00
# ibv_create_xrc_rcv_qp was added in OFED 1.3
2014-12-09 12:43:15 +03:00
# ibv_cmd_open_xrcd (aka XRC Domains) was added in OFED 3.12
2008-10-28 20:22:29 +03:00
if test "$enable_connectx_xrc" = "yes"; then
2015-07-13 03:56:51 +03:00
AC_CHECK_DECLS([IBV_DEVICE_XRC],
[$1_have_xrc=1
$1_have_xrc_domains=1],
[],
2015-07-10 16:44:40 +03:00
[#include <infiniband/verbs.h>])
fi
if test "$enable_connectx_xrc" = "yes" \
&& test $$1_have_xrc -eq 1; then
2015-07-13 03:56:51 +03:00
AC_CHECK_DECLS([ibv_create_xrc_rcv_qp],
[AC_CHECK_FUNCS([ibv_create_xrc_rcv_qp],
[], [$1_have_xrc=0])],
[$1_have_xrc=0],
[#include <infiniband/driver.h>])
fi
if test "$enable_connectx_xrc" = "yes" \
&& test $$1_have_xrc_domains -eq 1; then
AC_CHECK_DECLS([ibv_cmd_open_xrcd],
[AC_CHECK_DECLS([IBV_SRQT_XRC],
[AC_CHECK_FUNCS([ibv_cmd_open_xrcd],
[], [$1_have_xrc_domains=0])],
[$1_have_xrc_domains=0],
[#include <infiniband/verbs.h>])],
[$1_have_xrc_domains=0],
[#include <infiniband/driver.h>])
# XRC and XRC Domains should be considered as exclusive
if test "$$1_have_xrc" -eq 1 && \
test "$$1_have_xrc_domains" -eq 1; then
$1_have_xrc=0
fi
2008-10-28 20:22:29 +03:00
fi
2014-12-09 12:43:15 +03:00
2011-07-14 03:19:58 +04:00
if test "no" != "$enable_openib_dynamic_sl"; then
# We need ib_types.h file, which is installed with opensm-devel
# package. However, ib_types.h has a bad include directive,
# which will cause AC_CHECK_HEADER to fail.
# So instead, we will look for another file that is also
# installed as part of opensm-devel package and included in
# ib_types.h, but it doesn't include any other IB-related files.
AC_CHECK_HEADER([infiniband/complib/cl_types_osd.h],
2012-06-19 21:45:56 +04:00
[AC_CHECK_LIB([osmcomp], [cl_map_init],
[$1_have_opensm_devel=1],[])],
[],
[])
2011-07-14 03:19:58 +04:00
# Abort if dynamic SL support was explicitly requested but opensm-devel
# package wasn't found. Otherwise, OMPI will be built w/o dynamic SL.
AC_MSG_CHECKING([if can use dynamic SL support])
AS_IF([test "$$1_have_opensm_devel" = "1"],
[AC_MSG_RESULT([yes])],
[AC_MSG_RESULT([no])
AS_IF([test "$enable_openib_dynamic_sl" = "yes"],
[AC_MSG_WARN([--enable-openib-dynamic-sl was specified but the])
2012-06-19 21:45:56 +04:00
AC_MSG_WARN([appropriate header/library files could not be found])
2011-07-14 03:19:58 +04:00
AC_MSG_WARN([Please install opensm-devel if you need dynamic SL support])
AC_MSG_ERROR([Cannot continue])])])
fi
2011-06-28 18:28:29 +04:00
2008-10-28 20:22:29 +03:00
2009-12-15 16:52:19 +03:00
# Check support for RDMAoE devices
$1_have_rdmaoe=0
AC_CHECK_DECLS([IBV_LINK_LAYER_ETHERNET],
[$1_have_rdmaoe=1], [],
[#include <infiniband/verbs.h>])
AC_MSG_CHECKING([if RDMAoE support is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_HAVE_RDMAOE], [$$1_have_rdmaoe], [Enable RDMAoE support])
2009-12-15 16:52:19 +03:00
if test "1" = "$$1_have_rdmaoe"; then
AC_MSG_RESULT([yes])
else
AC_MSG_RESULT([no])
fi
2008-10-28 20:22:29 +03:00
])
2008-10-30 00:45:06 +03:00
# Check to see if <infiniband/driver.h> works. It is known to
# create problems on some platforms with some compilers (e.g.,
# RHEL4U3 with the PGI 32 bit compiler). Use undocumented (in AC
# 2.63) feature of AC_CHECK_HEADERS: if you explicitly pass in
# AC_INCLUDES_DEFAULT as the 4th arg to AC_CHECK_HEADERS, the test
# will fail if the header is present but not compilable, *but it
# will not print the big scary warning*. See
# http://lists.gnu.org/archive/html/autoconf/2008-10/msg00143.html.
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2015-06-24 06:59:57 +03:00
[AC_CHECK_HEADERS([infiniband/driver.h], [], [],
2008-10-30 00:45:06 +03:00
[AC_INCLUDES_DEFAULT])])
2008-10-28 20:22:29 +03:00
AC_MSG_CHECKING([if ConnectX XRC support is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_HAVE_CONNECTX_XRC], [$$1_have_xrc],
2008-10-28 20:22:29 +03:00
[Enable features required for ConnectX XRC support])
if test "1" = "$$1_have_xrc"; then
AC_MSG_RESULT([yes])
else
AC_MSG_RESULT([no])
fi
2011-06-28 18:28:29 +04:00
2014-12-09 12:43:15 +03:00
AC_MSG_CHECKING([if ConnectIB XRC support is enabled])
2015-01-07 07:27:25 +03:00
AC_DEFINE_UNQUOTED([OPAL_HAVE_CONNECTX_XRC_DOMAINS], [$$1_have_xrc_domains],
2014-12-09 12:43:15 +03:00
[Enable features required for XRC domains support])
2015-01-07 07:27:25 +03:00
if test "1" = "$$1_have_xrc_domains"; then
2014-12-09 12:43:15 +03:00
AC_MSG_RESULT([yes])
else
AC_MSG_RESULT([no])
fi
2011-07-14 03:19:58 +04:00
AC_MSG_CHECKING([if dynamic SL is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_ENABLE_DYNAMIC_SL], [$$1_have_opensm_devel],
2011-07-14 03:19:58 +04:00
[Enable features required for dynamic SL support])
if test "1" = "$$1_have_opensm_devel"; then
AC_MSG_RESULT([yes])
2012-06-19 21:45:56 +04:00
$1_LIBS="-losmcomp $$1_LIBS"
2011-07-14 03:19:58 +04:00
else
AC_MSG_RESULT([no])
2011-06-28 18:28:29 +04:00
fi
2011-07-03 18:02:50 +04:00
2012-07-02 19:20:12 +04:00
AS_IF([test -z "$opal_verbs_dir"],
[openib_include_dir="/usr/include"],
[openib_include_dir="$opal_verbs_dir/include"])
2013-01-07 22:57:03 +04:00
$1_CPPFLAGS="$$1_CPPFLAGS -I$openib_include_dir/infiniband"
2012-07-02 19:20:12 +04:00
2013-01-07 22:57:03 +04:00
CPPFLAGS="$ompi_check_openib_$1_save_CPPFLAGS"
2012-07-02 19:20:12 +04:00
LDFLAGS="$ompi_check_openib_$1_save_LDFLAGS"
LIBS="$ompi_check_openib_$1_save_LIBS"
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2012-07-02 19:20:12 +04:00
[$2],
[AS_IF([test "$opal_want_verbs" = "yes"],
[AC_MSG_WARN([Verbs support requested (via --with-verbs) but not found.])
AC_MSG_WARN([If you are using libibverbs v1.0 (i.e., OFED v1.0 or v1.1), you *MUST* have both the libsysfs headers and libraries installed. Later versions of libibverbs do not require libsysfs.])
AC_MSG_ERROR([Aborting.])])
$3])
OPAL_VAR_SCOPE_POP
])
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
AC_DEFUN([OPAL_CHECK_OPENFABRICS_CM_ARGS],[
2012-07-02 19:20:12 +04:00
#
# ConnectX XRC support
#
AC_ARG_ENABLE([openib-connectx-xrc],
[AC_HELP_STRING([--enable-openib-connectx-xrc],
[Enable ConnectX XRC support in the openib BTL. If you do not have InfiniBand ConnectX adapters, you may disable the ConnectX XRC support. If you do not know which InfiniBand adapter is installed on your cluster, leave this option enabled (default: enabled)])],
[enable_connectx_xrc="$enableval"], [enable_connectx_xrc="yes"])
#
# Unconnect Datagram (UD) based connection manager
#
2013-05-09 20:10:29 +04:00
AC_ARG_ENABLE([openib-udcm],
[AC_HELP_STRING([--enable-openib-udcm],
2015-06-24 06:59:57 +03:00
[Enable datagram connection support in openib BTL (default: enabled)])],
2013-05-09 20:10:29 +04:00
[enable_openib_udcm="$enableval"], [enable_openib_udcm="yes"])
2012-09-13 22:25:10 +04:00
# Per discussion with Ralph and Nathan, disable UDCM for now.
# It's borked and needs some surgery to get back on its feet.
2013-05-09 20:10:29 +04:00
# enable_openib_udcm=no
2013-01-07 21:54:10 +04:00
2012-07-02 19:20:12 +04:00
#
# Openfabrics RDMACM
#
AC_ARG_ENABLE([openib-rdmacm],
[AC_HELP_STRING([--enable-openib-rdmacm],
[Enable Open Fabrics RDMACM support in openib BTL (default: enabled)])])
2014-02-27 15:29:47 +04:00
AC_ARG_ENABLE([openib-rdmacm-ibaddr],
[AC_HELP_STRING([--enable-openib-rdmacm-ibaddr],
[Enable Open Fabrics RDMACM with IB addressing support in openib BTL (default: disabled)])],
[enable_openib_rdmacm=yes])
2012-07-02 19:20:12 +04:00
])dnl
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
AC_DEFUN([OPAL_CHECK_OPENFABRICS_CM],[
AC_REQUIRE([OPAL_CHECK_OPENFABRICS_CM_ARGS])
2012-07-02 19:20:12 +04:00
$1_have_udcm=0
$1_have_rdmacm=0
2013-01-07 21:54:10 +04:00
ompi_check_openib_$1_save_CPPFLAGS="$CPPFLAGS"
ompi_check_openib_$1_save_LDFLAGS="$LDFLAGS"
ompi_check_openib_$1_save_LIBS="$LIBS"
# add back in all the InfiniBand flags so that these tests might work...
CPPFLAGS="$CPPFLAGS $$1_CPPFLAGS"
LDFLAGS="$LDFLAGS $$1_LDFLAGS"
LIBS="$LIBS $$1_LIBS"
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2012-07-02 19:20:12 +04:00
[# Do we have a recent enough RDMA CM? Need to have the
# rdma_get_peer_addr (inline) function (originally appeared
# in OFED v1.3).
if test "$enable_openib_rdmacm" != "no"; then
AC_CHECK_HEADERS([rdma/rdma_cma.h],
[AC_CHECK_LIB([rdmacm], [rdma_create_id],
[AC_MSG_CHECKING([for rdma_get_peer_addr])
$1_msg=no
AC_LINK_IFELSE([AC_LANG_PROGRAM([[#include "rdma/rdma_cma.h"
]], [[void *ret = (void*) rdma_get_peer_addr((struct rdma_cm_id*)0);]])],
2015-06-24 06:59:57 +03:00
[$1_have_rdmacm=1
2012-07-02 19:20:12 +04:00
$1_msg=yes])
AC_MSG_RESULT([$$1_msg])])])
if test "1" = "$$1_have_rdmacm"; then
$1_LIBS="-lrdmacm $$1_LIBS"
else
AS_IF([test "$enable_openib_rdmacm" = "yes"],
[AC_MSG_WARN([--enable-openib-rdmacm was specified but the])
AC_MSG_WARN([appropriate files could not be found])
AC_MSG_WARN([Please install librdmacm and librdmacm-devel or disable rdmacm support])
AC_MSG_ERROR([Cannot continue])])
fi
fi
# is udcm enabled
if test "$enable_openib_udcm" = "yes"; then
$1_have_udcm=1
fi
])
2013-01-07 21:54:10 +04:00
CPPFLAGS="$ompi_check_openib_$1_save_CPPFLAGS"
LDFLAGS="$ompi_check_openib_$1_save_LDFLAGS"
LIBS="$ompi_check_openib_$1_save_LIBS"
2012-05-08 21:20:09 +04:00
AC_MSG_CHECKING([if UD CM is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_HAVE_UDCM], [$$1_have_udcm],
2012-04-03 15:02:34 +04:00
[Whether UD CM is available or not])
2011-09-20 01:24:58 +04:00
if test "1" = "$$1_have_udcm"; then
2012-04-03 15:02:34 +04:00
AC_MSG_RESULT([yes])
2011-09-20 01:24:58 +04:00
else
2012-04-03 15:02:34 +04:00
AC_MSG_RESULT([no])
2011-09-20 01:24:58 +04:00
fi
2008-10-28 20:22:29 +03:00
AC_MSG_CHECKING([if OpenFabrics RDMACM support is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_HAVE_RDMACM], [$$1_have_rdmacm],
2008-10-28 20:22:29 +03:00
[Whether RDMA CM is available or not])
if test "1" = "$$1_have_rdmacm"; then
AC_MSG_RESULT([yes])
else
AC_MSG_RESULT([no])
fi
2012-07-02 19:20:12 +04:00
])dnl
2008-10-28 20:22:29 +03:00
2015-11-24 02:07:12 +03:00
AC_DEFUN([OPAL_CHECK_EXP_VERBS],[
OPAL_VAR_SCOPE_PUSH([have_struct_ibv_exp_send_wr])
AC_MSG_CHECKING([whether expanded verbs are available])
AC_TRY_COMPILE([#include <infiniband/verbs_exp.h>], [struct ibv_exp_send_wr;],
[have_struct_ibv_exp_send_wr=1
AC_MSG_RESULT([yes])],
[have_struct_ibv_exp_send_wr=0
AC_MSG_RESULT([no])])
2015-11-30 22:32:04 +03:00
AC_DEFINE_UNQUOTED([HAVE_EXP_VERBS], [$have_struct_ibv_exp_send_wr], [Experimental verbs])
2015-12-10 03:02:36 +03:00
AC_CHECK_DECLS([IBV_EXP_ATOMIC_HCA_REPLY_BE, IBV_EXP_QP_CREATE_ATOMIC_BE_REPLY, ibv_exp_create_qp, ibv_exp_query_device, IBV_EXP_QP_INIT_ATTR_ATOMICS_ARG],
2015-11-30 22:32:04 +03:00
[], [], [#include <infiniband/verbs_exp.h>])
2015-11-24 02:07:12 +03:00
AS_IF([test '$have_struct_ibv_exp_send_wr' = 1], [$1], [$2])
OPAL_VAR_SCOPE_POP
])dnl
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
AC_DEFUN([OPAL_CHECK_MLNX_OPENFABRICS],[
2012-07-02 19:20:12 +04:00
$1_have_mverbs=0
$1_have_mqe=0
2011-06-28 18:28:29 +04:00
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2014-05-01 18:24:56 +04:00
[OPAL_CHECK_PACKAGE([$1],
2012-07-02 19:20:12 +04:00
[infiniband/mverbs.h],
[mverbs],
[ibv_m_query_device],
["$$1_LIBS"],
[$opal_verbs_dir],
[$opal_verbs_libdir],
[$1_have_mverbs=1],
[])])
2008-10-28 20:22:29 +03:00
2015-10-13 08:12:52 +03:00
AS_IF([test "$opal_check_openib_happy" = "yes"],
2014-05-01 18:24:56 +04:00
[OPAL_CHECK_PACKAGE([$1],
2012-07-02 19:20:12 +04:00
[infiniband/mqe.h],
[mqe],
[mqe_context_create],
["$$1_LIBS"],
[$opal_verbs_dir],
[$opal_verbs_libdir],
[$1_have_mqe=1],
[])])
AC_MSG_CHECKING([if Mellanox OpenFabrics VERBS is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_HAVE_MVERBS], [$$1_have_mverbs],
2012-07-02 19:20:12 +04:00
[Whether MVERBS is available or not])
AS_IF([test "1" = "$$1_have_mverbs"],
[AC_MSG_RESULT([yes])],
[AC_MSG_RESULT([no])])
# save the CPPFLAGS since we would have to update it for next test
ompi_check_mellanox_openfabrics_$1_save_CPPFLAGS="$CPPFLAGS"
# If openfabrics custom directory have been defined, we have
# to use it for MACRO test that uses mverbs.h file.
#
if test ! -z "$ompi_check_verbs_dir" ; then
CPPFLAGS="-I${opal_verbs_dir}/include $CPPFLAGS"
fi
2008-10-28 20:22:29 +03:00
2012-07-02 19:20:12 +04:00
AS_IF([test "1" = "$$1_have_mverbs"],
2015-06-24 06:59:57 +03:00
[AC_CHECK_DECLS([IBV_M_WR_CALC_RDMA_WRITE_WITH_IMM],
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
[AC_DEFINE_UNQUOTED([OPAL_HAVE_IBOFFLOAD_CALC_RDMA], [1],
2015-06-24 06:59:57 +03:00
[Whether IBV_M_WR_CALC_SEND is defined or not])],
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
[AC_DEFINE_UNQUOTED([OPAL_HAVE_IBOFFLOAD_CALC_RDMA], [0],
2015-06-24 06:59:57 +03:00
[Whether IBV_M_WR_CALC_SEND is defined or not])],
2012-07-02 19:20:12 +04:00
[#include <infiniband/mverbs.h>])])
# restoring the CPPFLAGS
CPPFLAGS="$ompi_check_mellanox_openfabrics_$1_save_CPPFLAGS"
AC_MSG_CHECKING([if Mellanox OpenFabrics MQE is enabled])
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
AC_DEFINE_UNQUOTED([OPAL_HAVE_MQE], [$$1_have_mqe],
2012-07-02 19:20:12 +04:00
[Whether MQE is available or not])
AS_IF([test "1" = "$$1_have_mqe"],
[AC_MSG_RESULT([yes])],
[AC_MSG_RESULT([no])])
2015-02-03 09:19:22 +03:00
AS_IF([test "1" = "$$1_have_mverbs" && test "1" = $$1_have_mqe],
2012-07-02 19:20:12 +04:00
[$2], [$3])
])dnl
