Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

PoC: Add groundwork for separating actual cuda parts in OpenMPI. WIP WIP WIP #14919

Closed
wants to merge 3 commits into from
Closed

PoC: Add groundwork for separating actual cuda parts in OpenMPI. WIP WIP WIP #14919

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ee2b250
Add groundwork for separating actual cuda parts in OpenMPI.
akesandgren Feb 3, 2022
1c75cb9
Add easyconfig for OpenMPI-CUDA. Depends on https://github.com/easybu…
akesandgren Apr 20, 2022
75396f0
add missing checksums
akesandgren Apr 20, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
351 changes: 351 additions & 0 deletions easybuild/easyconfigs/o/OpenMPI-CUDA/OpenMPI-4.0.6_remove-pmix-check-in-pmi-switch.patch
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,351 @@
If --with-pmi is given OpenMPI will always search for a PMIx in the same directory and the system first
before honoring a potential --with-pmix option.
This patch removes that.
See https://github.com/open-mpi/ompi/issues/9105

Author: Alexander Grund (TU Dresden)

diff -ur a/configure b/configure
--- a/configure 2021-06-14 22:53:22.000000000 +0200
+++ b/configure 2021-07-01 18:16:23.062505971 +0200
@@ -13441,340 +13441,6 @@
fi
fi

- # check for pmix lib installed by slurm */
- slurm_pmix_found=no
-
-
-
-
- # Is the private index set? If not, set it.
- if test "x$opal_scope_index" = "x"; then
- opal_scope_index=1
- fi
-
- # First, check to see if any of these variables are already set.
- # This is a simple sanity check to ensure we're not already
- # overwriting pre-existing variables (that have a non-empty
- # value). It's not a perfect check, but at least it's something.
- for opal_var in opal_external_pmix_save_CPPFLAGS opal_external_pmix_save_LDFLAGS opal_external_pmix_save_LIBS; do
- opal_str="opal_str=\"\$$opal_var\""
- eval $opal_str
-
- if test "x$opal_str" != "x"; then
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Found configure shell variable clash at line $LINENO!" >&5
-$as_echo "$as_me: WARNING: Found configure shell variable clash at line $LINENO!" >&2;}
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: OPAL_VAR_SCOPE_PUSH called on \"$opal_var\"," >&5
-$as_echo "$as_me: WARNING: OPAL_VAR_SCOPE_PUSH called on \"$opal_var\"," >&2;}
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: but it is already defined with value \"$opal_str\"" >&5
-$as_echo "$as_me: WARNING: but it is already defined with value \"$opal_str\"" >&2;}
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: This usually indicates an error in configure." >&5
-$as_echo "$as_me: WARNING: This usually indicates an error in configure." >&2;}
- as_fn_error $? "Cannot continue" "$LINENO" 5
- fi
- done
-
- # Ok, we passed the simple sanity check. Save all these names so
- # that we can unset them at the end of the scope.
- opal_str="opal_scope_$opal_scope_index=\"opal_external_pmix_save_CPPFLAGS opal_external_pmix_save_LDFLAGS opal_external_pmix_save_LIBS\""
- eval $opal_str
- unset opal_str
-
- env | grep opal_scope
- opal_scope_index=`expr $opal_scope_index + 1`
-
- opal_external_pmix_happy=no
-
- # Make sure we have the headers and libs in the correct location
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking for pmix.h in $check_pmi_install_dir" >&5
-$as_echo_n "checking for pmix.h in $check_pmi_install_dir... " >&6; }
- files=`ls $check_pmi_install_dir/pmix.h 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- opal_external_pmix_header_happy=yes
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking for pmix.h in $check_pmi_install_dir/include" >&5
-$as_echo_n "checking for pmix.h in $check_pmi_install_dir/include... " >&6; }
- files=`ls $check_pmi_install_dir/include/pmix.h 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- opal_external_pmix_header_happy=yes
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- opal_external_pmix_header_happy=no
-fi
-fi
-
- if test "$opal_external_pmix_header_happy" = "yes"; then :
- if test -n "$check_pmi_lib_dir"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking libpmix.* in $check_pmi_lib_dir" >&5
-$as_echo_n "checking libpmix.* in $check_pmi_lib_dir... " >&6; }
- files=`ls $check_pmi_lib_dir/libpmix.* 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- pmix_ext_install_libdir=$check_pmi_lib_dir
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking libpmix.* in $check_pmi_lib_dir/lib64" >&5
-$as_echo_n "checking libpmix.* in $check_pmi_lib_dir/lib64... " >&6; }
- files=`ls $check_pmi_lib_dir/lib64/libpmix.* 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- pmix_ext_install_libdir=$check_pmi_lib_dir/lib64
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking libpmix.* in $check_pmi_lib_dir/lib" >&5
-$as_echo_n "checking libpmix.* in $check_pmi_lib_dir/lib... " >&6; }
- files=`ls $check_pmi_lib_dir/lib/libpmix.* 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- pmix_ext_install_libdir=$check_pmi_lib_dir/lib
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- as_fn_error $? "Cannot continue" "$LINENO" 5
-fi
-fi
-fi
-else
- # check for presence of lib64 directory - if found, see if the
- # desired library is present and matches our build requirements
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking libpmix.* in $check_pmi_install_dir/lib64" >&5
-$as_echo_n "checking libpmix.* in $check_pmi_install_dir/lib64... " >&6; }
- files=`ls $check_pmi_install_dir/lib64/libpmix.* 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- pmix_ext_install_libdir=$check_pmi_install_dir/lib64
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking libpmix.* in $check_pmi_install_dir/lib" >&5
-$as_echo_n "checking libpmix.* in $check_pmi_install_dir/lib... " >&6; }
- files=`ls $check_pmi_install_dir/lib/libpmix.* 2> /dev/null | wc -l`
- if test "$files" -gt 0; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- pmix_ext_install_libdir=$check_pmi_install_dir/lib
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
- as_fn_error $? "Cannot continue" "$LINENO" 5
-fi
-fi
-fi
-
- # check the version
- opal_external_pmix_save_CPPFLAGS=$CPPFLAGS
- opal_external_pmix_save_LDFLAGS=$LDFLAGS
- opal_external_pmix_save_LIBS=$LIBS
-
- # if the pmix_version.h file does not exist, then
- # this must be from a pre-1.1.5 version OMPI does
- # NOT support anything older than v1.2.5
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking PMIx version" >&5
-$as_echo_n "checking PMIx version... " >&6; }
- CPPFLAGS="-I$check_pmi_install_dir/include $CPPFLAGS"
- if test "x`ls $check_pmi_install_dir/include/pmix_version.h 2> /dev/null`" = "x"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: version file not found - assuming v1.1.4" >&5
-$as_echo "version file not found - assuming v1.1.4" >&6; }
- opal_external_pmix_version_found=1
- opal_external_pmix_happy=no
- opal_external_pmix_version=internal
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: version file found" >&5
-$as_echo "version file found" >&6; }
- opal_external_pmix_version_found=0
-fi
-
- # if it does exist, then we need to parse it to find
- # the actual release series
- if test "$opal_external_pmix_version_found" = "0"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking version 4x" >&5
-$as_echo_n "checking version 4x... " >&6; }
- cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h. */
-
- #include <pmix_version.h>
- #if (PMIX_VERSION_MAJOR < 4L)
- #error "not version 4 or above"
- #endif
-
-int
-main ()
-{
-
- ;
- return 0;
-}
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- opal_external_pmix_version=4x
- opal_external_pmix_version_major=4
- opal_external_pmix_version_found=1
- opal_external_pmix_happy=yes
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-fi
-
- if test "$opal_external_pmix_version_found" = "0"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking version 3x or above" >&5
-$as_echo_n "checking version 3x or above... " >&6; }
- cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h. */
-
- #include <pmix_version.h>
- #if (PMIX_VERSION_MAJOR != 3L)
- #error "not version 3"
- #endif
-
-int
-main ()
-{
-
- ;
- return 0;
-}
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- opal_external_pmix_version=3x
- opal_external_pmix_version_major=3
- opal_external_pmix_version_found=1
- opal_external_pmix_happy=yes
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-fi
-
- if test "$opal_external_pmix_version_found" = "0"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking version 2x" >&5
-$as_echo_n "checking version 2x... " >&6; }
- cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h. */
-
- #include <pmix_version.h>
- #if (PMIX_VERSION_MAJOR != 2L)
- #error "not version 2"
- #endif
-
-int
-main ()
-{
-
- ;
- return 0;
-}
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- opal_external_pmix_version=2x
- opal_external_pmix_version_major=2
- opal_external_pmix_version_found=1
- opal_external_pmix_happy=yes
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-fi
-
- if test "$opal_external_pmix_version_found" = "0"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking version 1x" >&5
-$as_echo_n "checking version 1x... " >&6; }
- cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h. */
-
- #include <pmix_version.h>
- #if (PMIX_VERSION_MAJOR != 1L && PMIX_VERSION_MINOR != 2L)
- #error "not version 1.2.x"
- #endif
-
-int
-main ()
-{
-
- ;
- return 0;
-}
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: found" >&5
-$as_echo "found" >&6; }
- opal_external_pmix_version=1x
- opal_external_pmix_version_major=1
- opal_external_pmix_version_found=1
- opal_external_have_pmix1=1
- opal_external_pmix_happy=yes
-else
- { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-fi
-
- if test "x$opal_external_pmix_version" = "x"; then :
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: External PMIx support detected, but version" >&5
-$as_echo "$as_me: WARNING: External PMIx support detected, but version" >&2;}
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: information of the external lib could not" >&5
-$as_echo "$as_me: WARNING: information of the external lib could not" >&2;}
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: be detected" >&5
-$as_echo "$as_me: WARNING: be detected" >&2;}
- opal_external_pmix_happy=no
-fi
-
- CPPFLAGS=$opal_external_pmix_save_CPPFLAGS
- LDFLAGS=$opal_external_pmix_save_LDFLAGS
- LIBS=$opal_external_pmix_save_LIBS
-
-fi
- if test "$opal_external_pmix_happy" = "yes"; then :
- opal_enable_pmix=yes
-else
- opal_enable_pmix=no
-fi
-
-
- # Unwind the index
- opal_scope_index=`expr $opal_scope_index - 1`
- opal_scope_test=`expr $opal_scope_index \> 0`
- if test "$opal_scope_test" = "0"; then
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: OPAL_VAR_SCOPE_POP popped too many OPAL configure scopes." >&5
-$as_echo "$as_me: WARNING: OPAL_VAR_SCOPE_POP popped too many OPAL configure scopes." >&2;}
- { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: This usually indicates an error in configure." >&5
-$as_echo "$as_me: WARNING: This usually indicates an error in configure." >&2;}
- as_fn_error $? "Cannot continue" "$LINENO" 5
- fi
-
- # Get the variable names from that index
- opal_str="opal_str=\"\$opal_scope_$opal_scope_index\""
- eval $opal_str
-
- # Iterate over all the variables and unset them all
- for opal_var in $opal_str; do
- unset $opal_var
- done
-
-
-
# since support was explicitly requested, then we should error out
# if we didn't find the required support
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking can PMI support be built" >&5
Loading