btl/ugni: remove ugni btl and related doc notes

Note all references to uGNI are removed from the docs as
some are referenced via UCX support, etc.

We are removing this as we no longer have access to systems
using Cray legacy gemini and aries networks where uGNI was used.

An entry will be needed to the yet-to-be started changelog for v6.0.x

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

Author: hppritcha

VERSION

@@ -118,4 +118,3 @@ libmca_opal_common_cuda_so_version=0:0:0
 libmca_opal_common_ofi_so_version=0:0:0
 libmca_opal_common_sm_so_version=0:0:0
 libmca_opal_common_ucx_so_version=0:0:0
-libmca_opal_common_ugni_so_version=0:0:0

config/opal_check_ugni.m4

@@ -172,7 +172,6 @@ AC_SUBST(libopen_pal_so_version)
 AC_SUBST(libmca_opal_common_ofi_so_version)
 AC_SUBST(libmca_opal_common_cuda_so_version)
 AC_SUBST(libmca_opal_common_sm_so_version)
-AC_SUBST(libmca_opal_common_ugni_so_version)
 AC_SUBST(libmca_ompi_common_ompio_so_version)
 AC_SUBST(libmca_ompi_common_monitoring_so_version)
 AC_SUBST(libmca_opal_common_ucx_so_version)

configure.ac

@@ -41,8 +41,7 @@ can be used with ``configure``:
   compiler/linker search paths.
 
   Libfabric is the support library for OpenFabrics Interfaces-based
-  network adapters, such as Cisco usNIC, Intel True Scale PSM, Cray
-  uGNI, etc.
+  network adapters, such as Cisco usNIC, Intel True Scale PSM, etc.
 
 * ``--with-libfabric-libdir=DIR``:
   Look in directory for the libfabric libraries.  By default, Open MPI

docs/installing-open-mpi/configure-cli-options/networking.rst

@@ -156,7 +156,6 @@ Other MPI features
      #. ``sm``
      #. ``smcuda``
      #. ``tcp``
-     #. ``ugni``
      #. ``usnic``
 
   #. ``ucx``

docs/release-notes/mpi.rst

@@ -17,7 +17,6 @@ There are multiple MPI network models available in this release:
   * ``tcp``: TCP
   * ``smcuda``: CUDA-enabled shared memory
   * ``usnic``: Cisco usNIC
-  * ``ugni``: uGNI (Cray Gemini, Aries)
 
 * ``cm`` supports a smaller number of networks (and they cannot be
   used together), but may provide better overall MPI performance by
@@ -40,7 +39,6 @@ There are multiple MPI network models available in this release:
   UCX currently supports:
 
   * OpenFabrics Verbs (including InfiniBand and RoCE)
-  * Cray's uGNI
   * TCP
   * Shared memory
   * NVIDIA CUDA drivers
@@ -50,7 +48,7 @@ time, Open MPI will select a default transport as follows:
 
 #. If InfiniBand devices are available, use the UCX PML.
 #. If PSM, PSM2, or other tag-matching-supporting Libfabric
-   transport devices are available (e.g., Cray uGNI), use the ``cm``
+   transport devices are available, use the ``cm``
    PML and a single appropriate corresponding ``mtl`` module.
 #. Otherwise, use the ``ob1`` PML and one or more appropriate ``btl``
    modules.
@@ -104,8 +102,6 @@ Miscellaneous network notes
   Although the usNIC is accessed via the OpenFabrics Libfabric API
   stack, this BTL is specific to Cisco usNIC devices.
 
-* uGNI is a Cray library for communicating over the Gemini and Aries
-  interconnects.
 
 * Linux ``knem`` support is used when the ``sm`` (shared memory) BTL is
   compiled with knem support (see the ``--with-knem`` configure option)

docs/release-notes/networks.rst

@@ -28,7 +28,7 @@ Open MPI offers two flavors of CUDA support:
       # Check if ucx was built with CUDA support
       shell$ ucx_info -v
 
-      # configured with: --build=powerpc64le-redhat-linux-gnu --host=powerpc64le-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --disable-optimizations --disable-logging --disable-debug --disable-assertions --enable-mt --disable-params-check --enable-cma --without-cuda --without-gdrcopy --with-verbs --with-cm --with-knem --with-rdmacm --without-rocm --without-xpmem --without-ugni --without-java
+      # configured with: --build=powerpc64le-redhat-linux-gnu --host=powerpc64le-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --disable-optimizations --disable-logging --disable-debug --disable-assertions --enable-mt --disable-params-check --enable-cma --without-cuda --without-gdrcopy --with-verbs --with-cm --with-knem --with-rdmacm --without-rocm --without-xpmem --without-java
 
    If you need to build ucx yourself to include CUDA support, please
    see the UCX documentation for `building ucx with Open MPI: <https://openucx.readthedocs.io/en/master/running.html#openmpi-with-ucx>`_

docs/tuning-apps/networking/cuda.rst

@@ -16,7 +16,6 @@ including (but not limited to):
 
 * AWS EFA
 * Cisco usNIC
-* Cray uGNI
 * Cornelius Networks Omni-Path
 * HPE Slingshot 11
 

docs/tuning-apps/networking/ofi.rst

@@ -246,7 +246,7 @@ static int ompi_osc_rdma_component_register (void)
                                             MCA_BASE_VAR_SCOPE_GROUP, &mca_osc_rdma_component.locking_mode);
     OBJ_RELEASE(new_enum);
 
-    ompi_osc_rdma_full_connectivity_btls = "ugni,uct,ofi";
+    ompi_osc_rdma_full_connectivity_btls = "uct,ofi";
     opal_asprintf(&description_str, "Comma-delimited list of BTL component names to allow without verifying "
              "connectivity. Do not add a BTL to to this list unless it can reach all "
              "processes in any communicator used with an MPI window (default: %s)",

ompi/mca/osc/rdma/osc_rdma_component.c

@@ -5,7 +5,7 @@ enable RDMA with multiple vendor hardware through one interface. Most
 of the operations are managed by upper layer (osc/rdma). This BTL is
 mostly doing the low level work.
 
-Tested providers: sockets,psm2,ugni
+Tested providers: sockets,psm2
 
 ## Component