2015-02-03 09:19:22 +03:00
dnl -*- shell-script -*-
dnl
dnl Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
dnl University Research and Technology
dnl Corporation. All rights reserved.
dnl Copyright (c) 2004-2005 The University of Tennessee and The University
dnl of Tennessee Research Foundation. All rights
dnl reserved.
dnl Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
dnl University of Stuttgart. All rights reserved.
dnl Copyright (c) 2004-2006 The Regents of the University of California.
dnl All rights reserved.
dnl Copyright (c) 2006 QLogic Corp. All rights reserved.
dnl Copyright (c) 2009 Cisco Systems, Inc. All rights reserved.
dnl Copyright (c) 2015 Research Organization for Information Science
dnl and Technology (RIST). All rights reserved.
dnl $COPYRIGHT$
dnl
dnl Additional copyrights may follow
dnl
dnl $HEADER$
dnl
2010-10-15 02:49:09 +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
# OPAL_CHECK_PORTALS4(prefix, [action-if-found], [action-if-not-found])
2010-10-15 02:49:09 +04:00
# --------------------------------------------------------
2015-02-03 09:19:22 +03:00
# check if PORTALS4 support can be found. sets prefix_{CPPFLAGS,
2010-10-15 02:49:09 +04:00
# 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_PORTALS4],[
2010-10-15 02:49:09 +04:00
AC_ARG_WITH([portals4],
[AC_HELP_STRING([--with-portals4(=DIR)],
[Build Portals4 support, optionally adding DIR/include, DIR/lib, and DIR/lib64 to the search path for headers and libraries])])
2014-05-01 19:38:07 +04:00
OPAL_CHECK_WITHDIR([portals4], [$with_portals4], [include/portals4.h])
2010-10-15 02:49:09 +04:00
AC_ARG_WITH([portals4-libdir],
[AC_HELP_STRING([--with-portals4-libdir=DIR],
[Search for Portals4 libraries in DIR])])
2014-05-01 19:38:07 +04:00
OPAL_CHECK_WITHDIR([portals4-libdir], [$with_portals4_libdir], [libportals.*])
2010-10-15 02:49:09 +04:00
ompi_check_portals4_$1_save_CPPFLAGS="$CPPFLAGS"
ompi_check_portals4_$1_save_LDFLAGS="$LDFLAGS"
ompi_check_portals4_$1_save_LIBS="$LIBS"
AS_IF([test "$with_portals4" != "no"],
2015-02-03 09:19:22 +03:00
[AS_IF([test ! -z "$with_portals4" && test "$with_portals4" != "yes"],
2010-10-15 02:49:09 +04:00
[ompi_check_portals4_dir="$with_portals4"])
2015-02-03 09:19:22 +03:00
AS_IF([test ! -z "$with_portals4_libdir" && test "$with_portals4_libdir" != "yes"],
2010-10-15 02:49:09 +04:00
[ompi_check_portals4_libdir="$with_portals4_libdir"])
2014-05-01 18:24:56 +04:00
OPAL_CHECK_PACKAGE([$1],
2010-10-15 02:49:09 +04:00
[portals4.h],
[portals],
[PtlLEAppend],
[],
[$ompi_check_portals4_dir],
[$ompi_check_portals4_libdir],
[ompi_check_portals4_happy="yes"],
[ompi_check_portals4_happy="no"])],
[ompi_check_portals4_happy="no"])
CPPFLAGS="$ompi_check_portals4_$1_save_CPPFLAGS"
LDFLAGS="$ompi_check_portals4_$1_save_LDFLAGS"
LIBS="$ompi_check_portals4_$1_save_LIBS"
2013-07-09 01:27:37 +04:00
max_md_size=0
AC_ARG_WITH([portals4-max-md-size],
[AC_HELP_STRING([--with-portals4-max-md-size=SIZE],
[Log base 2 of the maximum size in bytes of a memory descriptor. Should only be set for implementations which do not support binding all of virtual address space.])])
2015-02-03 09:19:22 +03:00
AS_IF([test "$with_portals4_max_md_size" = "yes" || test "$with_portals4_max_md_size" = "no"],
2013-07-09 01:27:37 +04:00
[AC_MSG_ERROR([--with-portals4-max-md-size requires an integer argument])],
[AS_IF([test -n "$with_portals4_max_md_size"],
[max_md_size="$with_portals4_max_md_size"])])
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 04:47:28 +04:00
AC_DEFINE_UNQUOTED([OPAL_PORTALS4_MAX_MD_SIZE], [$max_md_size],
2013-07-09 01:27:37 +04:00
[Log base 2 of the maximum size in bytes of a memory descriptor. Set to 0 if MD can bind all of memory.])
max_va_size=0
AC_ARG_WITH([portals4-max-va-size],
[AC_HELP_STRING([--with-portals4-max-va-size=SIZE],
[Log base 2 of the maximum size in bytes of the user virtual address space. Should only be set for implementations which do not support binding all of virtual address space.])])
2015-02-03 09:19:22 +03:00
AS_IF([test "$with_portals4_max_va_size" = "yes" || test "$with_portals4_max_va_size" = "no"],
2013-07-09 01:27:37 +04:00
[AC_MSG_ERROR([--with-portals4-max-va-size requires an integer argument])],
[AS_IF([test -n "$with_portals4_max_va_size"],
[max_va_size="$with_portals4_max_va_size"])])
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_PORTALS4_MAX_VA_SIZE], [$max_va_size],
2013-07-09 01:27:37 +04:00
[Log base 2 of the maximum size in bytes of the user virtual address space. Set to 0 if MD can bind all of memory.])
2015-02-03 09:19:22 +03:00
AS_IF([(test $max_md_size -eq 0 && test $max_va_size -ne 0 ) || (test $max_md_size -ne 0 && test $max_va_size -eq 0 )],
2013-07-09 01:27:37 +04:00
[AC_ERROR([If either --with-portals4-max-md-size or --with-portals4-max-va-size is set, both must be set.])])
AS_IF([test $max_md_size -ge $max_va_size],
[max_md_size=0
max_va_size=0])
2015-02-03 09:19:22 +03:00
AS_IF([test $max_md_size -ne 0 && test $max_va_size -ne 0],
2013-07-09 01:27:37 +04:00
[AC_MSG_NOTICE([Portals 4 address space size: $max_md_size, $max_va_size])])
2010-10-15 02:49:09 +04:00
AS_IF([test "$ompi_check_portals4_happy" = "yes"],
[$2],
2015-02-03 09:19:22 +03:00
[AS_IF([test ! -z "$with_portals4" && test "$with_portals4" != "no"],
2010-10-15 02:49:09 +04:00
[AC_MSG_ERROR([Portals4 support requested but not found. Aborting])])
$3])
])