X-Git-Url: http://bilbo.iut-bm.univ-fcomte.fr/pub/gitweb/simgrid.git/blobdiff_plain/08f744b9a55745ac1b1dcf0ed2ea735471cd7f89..f973433226af2e2e3059b62578bb0f3b215380fb:/docs/source/app_smpi.rst?ds=sidebyside

diff --git a/docs/source/app_smpi.rst b/docs/source/app_smpi.rst
index 775fce95c8..a168562f74 100644
--- a/docs/source/app_smpi.rst
+++ b/docs/source/app_smpi.rst
@@ -6,7 +6,7 @@ SMPI: Simulate MPI Applications
 
 .. raw:: html
 
-   <object id="TOC" data="graphical-toc.svg" width="100%" type="image/svg+xml"></object>
+   <object id="TOC" data="graphical-toc.svg" type="image/svg+xml"></object>
    <script>
    window.onload=function() { // Wait for the SVG to be loaded before changing it
      var elem=document.querySelector("#TOC").contentDocument.getElementById("SMPIBox")
@@ -42,11 +42,121 @@ For **further scalability**, you may modify your code to speed up your
 studies or save memory space.  Maximal **simulation accuracy**
 requires some specific care from you.
 
+-------------------------
+What can run within SMPI?
+-------------------------
+
+You can run unmodified MPI applications (both C/C++ and Fortran) within
+SMPI, provided that you only use MPI calls that we implemented. Global
+variables should be handled correctly on Linux systems.
+
+....................
+MPI coverage of SMPI
+....................
+
+SMPI support a large faction of the MPI interface: we pass many of the MPICH coverage tests, and many of the existing
+:ref:`proxy apps <SMPI_proxy_apps>` run almost unmodified on top of SMPI. But our support is still incomplete, with I/O
+primitives the being one of the major missing feature.
+
+The full list of functions that remain to be implemented is documented in the file `include/smpi/smpi.h
+<https://framagit.org/simgrid/simgrid/tree/master/include/smpi/smpi.h>`_ in your version of SimGrid, between two lines
+containing the ``FIXME`` marker. If you miss a feature, please get in touch with us: we can guide you through the SimGrid
+code to help you implementing it, and we'd be glad to integrate your contribution to the main project.
+
+.. _SMPI_what_globals:
+
+.................................
+Privatization of global variables
+.................................
+
+Concerning the globals, the problem comes from the fact that usually,
+MPI processes run as real UNIX processes while they are all folded
+into threads of a unique system process in SMPI. Global variables are
+usually private to each MPI process while they become shared between
+the processes in SMPI.  The problem and some potential solutions are
+discussed in this article: `Automatic Handling of Global Variables for
+Multi-threaded MPI Programs
+<http://charm.cs.illinois.edu/newPapers/11-23/paper.pdf>` (note that
+this article does not deal with SMPI but with a competing solution
+called AMPI that suffers of the same issue).  This point used to be
+problematic in SimGrid, but the problem should now be handled
+automatically on Linux.
+
+Older versions of SimGrid came with a script that automatically
+privatized the globals through static analysis of the source code. But
+our implementation was not robust enough to be used in production, so
+it was removed at some point. Currently, SMPI comes with two
+privatization mechanisms that you can :ref:`select at runtime
+<cfg=smpi/privatization>`.  The dlopen approach is used by
+default as it is much faster and still very robust.  The mmap approach
+is an older approach that proves to be slower.
+
+With the **mmap approach**, SMPI duplicates and dynamically switch the
+``.data`` and ``.bss`` segments of the ELF process when switching the
+MPI ranks. This allows each ranks to have its own copy of the global
+variables.  No copy actually occurs as this mechanism uses ``mmap()``
+for efficiency. This mechanism is considered to be very robust on all
+systems supporting ``mmap()`` (Linux and most BSDs). Its performance
+is questionable since each context switch between MPI ranks induces
+several syscalls to change the ``mmap`` that redirects the ``.data``
+and ``.bss`` segments to the copies of the new rank. The code will
+also be copied several times in memory, inducing a slight increase of
+memory occupation.
+
+Another limitation is that SMPI only accounts for global variables
+defined in the executable. If the processes use external global
+variables from dynamic libraries, they won't be switched
+correctly. The easiest way to solve this is to statically link against
+the library with these globals. This way, each MPI rank will get its
+own copy of these libraries. Of course you should never statically
+link against the SimGrid library itself.
+
+With the **dlopen approach**, SMPI loads several copies of the same
+executable in memory as if it were a library, so that the global
+variables get naturally duplicated. It first requires the executable
+to be compiled as a relocatable binary, which is less common for
+programs than for libraries. But most distributions are now compiled
+this way for security reason as it allows one to randomize the address
+space layout. It should thus be safe to compile most (any?) program
+this way.  The second trick is that the dynamic linker refuses to link
+the exact same file several times, be it a library or a relocatable
+executable. It makes perfectly sense in the general case, but we need
+to circumvent this rule of thumb in our case. To that extend, the
+binary is copied in a temporary file before being re-linked against.
+``dlmopen()`` cannot be used as it only allows 256 contexts, and as it
+would also duplicate SimGrid itself.
+
+This approach greatly speeds up the context switching, down to about
+40 CPU cycles with our raw contexts, instead of requesting several
+syscalls with the ``mmap()`` approach. Another advantage is that it
+permits one to run the SMPI contexts in parallel, which is obviously not
+possible with the ``mmap()`` approach. It was tricky to implement, but
+we are not aware of any flaws, so smpirun activates it by default.
+
+In the future, it may be possible to further reduce the memory and
+disk consumption. It seems that we could `punch holes
+<https://lwn.net/Articles/415889/>`_ in the files before dl-loading
+them to remove the code and constants, and mmap these area onto a
+unique copy. If done correctly, this would reduce the disk- and
+memory- usage to the bare minimum, and would also reduce the pressure
+on the CPU instruction cache. See the `relevant bug
+<https://github.com/simgrid/simgrid/issues/137>`_ on github for
+implementation leads.\n
+
+Also, currently, only the binary is copied and dlopen-ed for each MPI
+rank. We could probably extend this to external dependencies, but for
+now, any external dependencies must be statically linked into your
+application. As usual, SimGrid itself shall never be statically linked
+in your app. You don't want to give a copy of SimGrid to each MPI rank:
+that's ways too much for them to deal with.
+
+.. todo: speak of smpi/privatize-libs here
+
 .. _SMPI_online:
 
------------------
-Using SMPI online
------------------
+---------------------------
+Online SMPI: live execution
+---------------------------
 
 In this mode, your application is actually executed. Every computation
 occurs for real while every communication is simulated. In addition,
@@ -74,9 +184,9 @@ If you use cmake, set the variables ``MPI_C_COMPILER``, ``MPI_CXX_COMPILER`` and
 ``MPI_Fortran_COMPILER`` to the full path of smpicc, smpicxx and smpiff (or
 smpif90), respectively. Example:
 
-.. code-block:: shell
+.. code-block:: console
 
-   cmake -DMPI_C_COMPILER=/opt/simgrid/bin/smpicc -DMPI_CXX_COMPILER=/opt/simgrid/bin/smpicxx -DMPI_Fortran_COMPILER=/opt/simgrid/bin/smpiff .
+   $ cmake -DMPI_C_COMPILER=/opt/simgrid/bin/smpicc -DMPI_CXX_COMPILER=/opt/simgrid/bin/smpicxx -DMPI_Fortran_COMPILER=/opt/simgrid/bin/smpiff .
 
 ....................
 Simulating your Code
@@ -84,13 +194,15 @@ Simulating your Code
 
 Use the ``smpirun`` script as follows:
 
-.. code-block:: shell
+.. code-block:: console
 
-   smpirun -hostfile my_hostfile.txt -platform my_platform.xml ./program -blah
+   $ smpirun -hostfile my_hostfile.txt -platform my_platform.xml ./program -blah
 
 - ``my_hostfile.txt`` is a classical MPI hostfile (that is, this file
   lists the machines on which the processes must be dispatched, one
-  per line)
+  per line). Using the ``hostname:num_procs`` syntax will deploy num_procs
+  MPI processes on the host, sharing available cores (equivalent to listing
+  the same host num_procs times on different lines).
 - ``my_platform.xml`` is a classical SimGrid platform file. Of course,
   the hosts of the hostfile must exist in the provided platform.
 - ``./program`` is the MPI program to simulate, that you compiled with ``smpicc``
@@ -102,6 +214,11 @@ on which host each rank gets mapped of ``-trace`` to activate the
 tracing during the simulation. You can get the full list by running
 ``smpirun -help``
 
+Finally, you can pass :ref:`any valid SimGrid parameter <options>` to your
+program. In particular, you can pass ``--cfg=network/model:ns-3`` to
+switch to use :ref:`models_ns3`. These parameters should be placed after
+the name of your binary on the command line.
+
 ...............................
 Debugging your Code within SMPI
 ...............................
@@ -115,10 +232,34 @@ following commands. Once in GDB, each MPI ranks will be represented as
 a regular thread, and you can explore the state of each of them as
 usual.
 
-.. code-block:: shell
+.. code-block:: console
+
+   $ smpirun -wrapper valgrind ...other args...
+   $ smpirun -wrapper "gdb --args" --cfg=contexts/factory:thread ...other args...
+
+Some shortcuts are available:
+
+- ``-gdb`` is equivalent to ``-wrapper "gdb --args" -keep-temps``, to run within gdb debugger
+- ``-lldb`` is equivalent to ``-wrapper "lldb --" -keep-temps``, to run within lldb debugger
+- ``-vgdb`` is equivalent to ``-wrapper "valgrind --vgdb=yes --vgdb-error=0" -keep-temps``,
+  to run within valgrind and allow to attach a debugger
+
+To help locate bottlenecks and largest allocations in the simulated application,
+the -analyze flag can be passed to smpirun. It will activate
+:ref:`smpi/display-timing<cfg=smpi/display-timing>` and
+:ref:`smpi/display-allocs<cfg=smpi/display-allocs>` options and provide hints
+at the end of execution.
+
+SMPI will also report MPI handle (Comm, Request, Op, Datatype...) leaks
+at the end of execution. This can help identify memory leaks that can trigger
+crashes and slowdowns.
+By default it only displays the number of leaked items detected.
+Option :ref:`smpi/list-leaks:n<cfg=smpi/list-leaks>` can be used to display the
+n first leaks encountered and their type. To get more information, running smpirun
+with ``-wrapper "valgrind --leak-check=full --track-origins=yes"`` should show
+the exact origin of leaked handles.
+Known issue : MPI_Cancel may trigger internal leaks within SMPI.
 
-   smpirun -wrapper valgrind ...other args...
-   smpirun -wrapper "gdb --args" --cfg=contexts/factory:thread ...other args...
 
 .. _SMPI_use_colls:
 
@@ -146,7 +287,7 @@ of the targeted MPI implementations.
 You can switch the automatic selector through the
 ``smpi/coll-selector`` configuration item. Possible values:
 
- - **ompi:** default selection logic of OpenMPI (version 3.1.2)
+ - **ompi:** default selection logic of OpenMPI (version 4.1.2)
  - **mpich**: default selection logic of MPICH (version 3.3b)
  - **mvapich2**: selection logic of MVAPICH2 (version 1.9) tuned
    on the Stampede cluster
@@ -172,265 +313,242 @@ means that the selected algorithm will be used
 .. Warning:: Some collective may require specific conditions to be
    executed correctly (for instance having a communicator with a power
    of two number of nodes only), which are currently not enforced by
-   Simgrid. Some crashes can be expected while trying these algorithms
+   SimGrid. Some crashes can be expected while trying these algorithms
    with unusual sizes/parameters
 
+To retrieve the full list of implemented algorithms in your version of SimGrid, simply use ``smpirun --help-coll``.
+
 MPI_Alltoall
 ^^^^^^^^^^^^
 
 Most of these are best described in `STAR-MPI's white paper <https://doi.org/10.1145/1183401.1183431>`_.
 
- - default: naive one, by default
- - ompi: use openmpi selector for the alltoall operations
- - mpich: use mpich selector for the alltoall operations
- - mvapich2: use mvapich2 selector for the alltoall operations
- - impi: use intel mpi selector for the alltoall operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - bruck: Described by Bruck et.al. in <a href="http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=642949">this paper</a>
- - 2dmesh: organizes the nodes as a two dimensional mesh, and perform allgather
-   along the dimensions
- - 3dmesh: adds a third dimension to the previous algorithm
- - rdb: recursive doubling: extends the mesh to a nth dimension, each one
-   containing two nodes
- - pair: pairwise exchange, only works for power of 2 procs, size-1 steps,
-   each process sends and receives from the same process at each step
- - pair_light_barrier: same, with small barriers between steps to avoid
-   contention
- - pair_mpi_barrier: same, with MPI_Barrier used
- - pair_one_barrier: only one barrier at the beginning
- - ring: size-1 steps, at each step a process send to process (n+i)%size, and receives from (n-i)%size
- - ring_light_barrier: same, with small barriers between some phases to avoid contention
- - ring_mpi_barrier: same, with MPI_Barrier used
- - ring_one_barrier: only one barrier at the beginning
- - basic_linear: posts all receives and all sends,
-   starts the communications, and waits for all communication to finish
- - mvapich2_scatter_dest: isend/irecv with scattered destinations, posting only a few messages at the same time
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the alltoall operations. |br|
+``mpich``: use mpich selector for the alltoall operations. |br|
+``mvapich2``: use mvapich2 selector for the alltoall operations. |br|
+``impi``: use intel mpi selector for the alltoall operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``bruck``: Described by Bruck et. al. in `this paper <http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=642949>`_. |br|
+``2dmesh``: organizes the nodes as a two dimensional mesh, and perform allgather along the dimensions. |br|
+``3dmesh``: adds a third dimension to the previous algorithm. |br|
+``rdb``: recursive doubling``: extends the mesh to a nth dimension, each one containing two nodes. |br|
+``pair``: pairwise exchange, only works for power of 2 procs, size-1 steps, each process sends and receives from the same process at each step. |br|
+``pair_light_barrier``: same, with small barriers between steps to avoid contention. |br|
+``pair_mpi_barrier``: same, with MPI_Barrier used. |br|
+``pair_one_barrier``: only one barrier at the beginning. |br|
+``ring``: size-1 steps, at each step a process send to process (n+i)%size, and receives from (n-i)%size. |br|
+``ring_light_barrier``: same, with small barriers between some phases to avoid contention. |br|
+``ring_mpi_barrier``: same, with MPI_Barrier used. |br|
+``ring_one_barrier``: only one barrier at the beginning. |br|
+``basic_linear``: posts all receives and all sends, starts the communications, and waits for all communication to finish. |br|
+``mvapich2_scatter_dest``: isend/irecv with scattered destinations, posting only a few messages at the same time. |br|
 
 MPI_Alltoallv
 ^^^^^^^^^^^^^
- - default: naive one, by default
- - ompi: use openmpi selector for the alltoallv operations
- - mpich: use mpich selector for the alltoallv operations
- - mvapich2: use mvapich2 selector for the alltoallv operations
- - impi: use intel mpi selector for the alltoallv operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - bruck: same as alltoall
- - pair: same as alltoall
- - pair_light_barrier: same as alltoall
- - pair_mpi_barrier: same as alltoall
- - pair_one_barrier: same as alltoall
- - ring: same as alltoall
- - ring_light_barrier: same as alltoall
- - ring_mpi_barrier: same as alltoall
- - ring_one_barrier: same as alltoall
- - ompi_basic_linear: same as alltoall
+
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the alltoallv operations. |br|
+``mpich``: use mpich selector for the alltoallv operations. |br|
+``mvapich2``: use mvapich2 selector for the alltoallv operations. |br|
+``impi``: use intel mpi selector for the alltoallv operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``bruck``: same as alltoall. |br|
+``pair``: same as alltoall. |br|
+``pair_light_barrier``: same as alltoall. |br|
+``pair_mpi_barrier``: same as alltoall. |br|
+``pair_one_barrier``: same as alltoall. |br|
+``ring``: same as alltoall. |br|
+``ring_light_barrier``: same as alltoall. |br|
+``ring_mpi_barrier``: same as alltoall. |br|
+``ring_one_barrier``: same as alltoall. |br|
+``ompi_basic_linear``: same as alltoall. |br|
 
 MPI_Gather
 ^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the gather operations
- - mpich: use mpich selector for the gather operations
- - mvapich2: use mvapich2 selector for the gather operations
- - impi: use intel mpi selector for the gather operations
- - automatic (experimental): use an automatic self-benchmarking algorithm which will iterate over all implemented versions and output the best
- - ompi_basic_linear: basic linear algorithm from openmpi, each process sends to the root
- - ompi_binomial: binomial tree algorithm
- - ompi_linear_sync: same as basic linear, but with a synchronization at the
-   beginning and message cut into two segments.
- - mvapich2_two_level: SMP-aware version from MVAPICH. Gather first intra-node (defaults to mpich's gather), and then exchange with only one process/node. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster.
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the gather operations. |br|
+``mpich``: use mpich selector for the gather operations. |br|
+``mvapich2``: use mvapich2 selector for the gather operations. |br|
+``impi``: use intel mpi selector for the gather operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm which will iterate over all implemented versions and output the best. |br|
+``ompi_basic_linear``: basic linear algorithm from openmpi, each process sends to the root. |br|
+``ompi_binomial``: binomial tree algorithm. |br|
+``ompi_linear_sync``: same as basic linear, but with a synchronization at the beginning and message cut into two segments. |br|
+``mvapich2_two_level``: SMP-aware version from MVAPICH. Gather first intra-node (defaults to mpich's gather), and then exchange with only one process/node. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster. |br|
 
 MPI_Barrier
 ^^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the barrier operations
- - mpich: use mpich selector for the barrier operations
- - mvapich2: use mvapich2 selector for the barrier operations
- - impi: use intel mpi selector for the barrier operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - ompi_basic_linear: all processes send to root
- - ompi_two_procs: special case for two processes
- - ompi_bruck: nsteps = sqrt(size), at each step, exchange data with rank-2^k and rank+2^k
- - ompi_recursivedoubling: recursive doubling algorithm
- - ompi_tree: recursive doubling type algorithm, with tree structure
- - ompi_doublering: double ring algorithm
- - mvapich2_pair: pairwise algorithm
- - mpich_smp: barrier intra-node, then inter-node
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the barrier operations. |br|
+``mpich``: use mpich selector for the barrier operations. |br|
+``mvapich2``: use mvapich2 selector for the barrier operations. |br|
+``impi``: use intel mpi selector for the barrier operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``ompi_basic_linear``: all processes send to root. |br|
+``ompi_two_procs``: special case for two processes. |br|
+``ompi_bruck``: nsteps = sqrt(size), at each step, exchange data with rank-2^k and rank+2^k. |br|
+``ompi_recursivedoubling``: recursive doubling algorithm. |br|
+``ompi_tree``: recursive doubling type algorithm, with tree structure. |br|
+``ompi_doublering``: double ring algorithm. |br|
+``mvapich2_pair``: pairwise algorithm. |br|
+``mpich_smp``: barrier intra-node, then inter-node. |br|
 
 MPI_Scatter
 ^^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the scatter operations
- - mpich: use mpich selector for the scatter operations
- - mvapich2: use mvapich2 selector for the scatter operations
- - impi: use intel mpi selector for the scatter operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - ompi_basic_linear: basic linear scatter
- - ompi_binomial: binomial tree scatter
- - mvapich2_two_level_direct: SMP aware algorithm, with an intra-node stage (default set to mpich selector), and then a basic linear inter node stage. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster. 
- - mvapich2_two_level_binomial: SMP aware algorithm, with an intra-node stage (default set to mpich selector), and then a binomial phase. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster.
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the scatter operations. |br|
+``mpich``: use mpich selector for the scatter operations. |br|
+``mvapich2``: use mvapich2 selector for the scatter operations. |br|
+``impi``: use intel mpi selector for the scatter operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``ompi_basic_linear``: basic linear scatter. |br|
+``ompi_linear_nb``: linear scatter, non blocking sends. |br|
+``ompi_binomial``: binomial tree scatter. |br|
+``mvapich2_two_level_direct``: SMP aware algorithm, with an intra-node stage (default set to mpich selector), and then a basic linear inter node stage. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster. |br|
+``mvapich2_two_level_binomial``: SMP aware algorithm, with an intra-node stage (default set to mpich selector), and then a binomial phase. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster. |br|
 
 MPI_Reduce
 ^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the reduce operations
- - mpich: use mpich selector for the reduce operations
- - mvapich2: use mvapich2 selector for the reduce operations
- - impi: use intel mpi selector for the reduce operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - arrival_pattern_aware: root exchanges with the first process to arrive
- - binomial: uses a binomial tree
- - flat_tree: uses a flat tree
- - NTSL: Non-topology-specific pipelined linear-bcast function
-   0->1, 1->2 ,2->3, ....., ->last node: in a pipeline fashion, with segments
-   of 8192 bytes
- - scatter_gather: scatter then gather
- - ompi_chain: openmpi reduce algorithms are built on the same basis, but the
-   topology is generated differently for each flavor
-   chain = chain with spacing of size/2, and segment size of 64KB
- - ompi_pipeline: same with pipeline (chain with spacing of 1), segment size
-   depends on the communicator size and the message size
- - ompi_binary: same with binary tree, segment size of 32KB
- - ompi_in_order_binary: same with binary tree, enforcing order on the
-   operations
- - ompi_binomial: same with binomial algo (redundant with default binomial
-   one in most cases)
- - ompi_basic_linear: basic algorithm, each process sends to root
- - mvapich2_knomial: k-nomial algorithm. Default factor is 4 (mvapich2 selector adapts it through tuning)
- - mvapich2_two_level: SMP-aware reduce, with default set to mpich both for intra and inter communicators. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster.
- - rab: `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_'s reduce algorithm
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the reduce operations. |br|
+``mpich``: use mpich selector for the reduce operations. |br|
+``mvapich2``: use mvapich2 selector for the reduce operations. |br|
+``impi``: use intel mpi selector for the reduce operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``arrival_pattern_aware``: root exchanges with the first process to arrive. |br|
+``binomial``: uses a binomial tree. |br|
+``flat_tree``: uses a flat tree. |br|
+``NTSL``: Non-topology-specific pipelined linear-bcast function. |br| 0->1, 1->2 ,2->3, ....., ->last node: in a pipeline fashion, with segments of 8192 bytes. |br|
+``scatter_gather``: scatter then gather. |br|
+``ompi_chain``: openmpi reduce algorithms are built on the same basis, but the topology is generated differently for each flavor. chain = chain with spacing of size/2, and segment size of 64KB. |br|
+``ompi_pipeline``: same with pipeline (chain with spacing of 1), segment size depends on the communicator size and the message size. |br|
+``ompi_binary``: same with binary tree, segment size of 32KB. |br|
+``ompi_in_order_binary``: same with binary tree, enforcing order on the operations. |br|
+``ompi_binomial``: same with binomial algo (redundant with default binomial one in most cases). |br|
+``ompi_basic_linear``: basic algorithm, each process sends to root. |br|
+``mvapich2_knomial``: k-nomial algorithm. Default factor is 4 (mvapich2 selector adapts it through tuning). |br|
+``mvapich2_two_level``: SMP-aware reduce, with default set to mpich both for intra and inter communicators. Use mvapich2 selector to change these to tuned algorithms for Stampede cluster. |br|
+``rab``: `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_'s reduce algorithm. |br|
 
 MPI_Allreduce
 ^^^^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the allreduce operations
- - mpich: use mpich selector for the allreduce operations
- - mvapich2: use mvapich2 selector for the allreduce operations
- - impi: use intel mpi selector for the allreduce operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - lr: logical ring reduce-scatter then logical ring allgather
- - rab1: variations of the  <a href="https://fs.hlrs.de/projects/par/mpi//myreduce.html">Rabenseifner</a> algorithm: reduce_scatter then allgather
- - rab2: variations of the  <a href="https://fs.hlrs.de/projects/par/mpi//myreduce.html">Rabenseifner</a> algorithm: alltoall then allgather
- - rab_rsag: variation of the  <a href="https://fs.hlrs.de/projects/par/mpi//myreduce.html">Rabenseifner</a> algorithm: recursive doubling
-   reduce_scatter then recursive doubling allgather
- - rdb: recursive doubling
- - smp_binomial: binomial tree with smp: binomial intra
-   SMP reduce, inter reduce, inter broadcast then intra broadcast
- - smp_binomial_pipeline: same with segment size = 4096 bytes
- - smp_rdb: intra: binomial allreduce, inter: Recursive
-   doubling allreduce, intra: binomial broadcast
- - smp_rsag: intra: binomial allreduce, inter: reduce-scatter,
-   inter:allgather, intra: binomial broadcast
- - smp_rsag_lr: intra: binomial allreduce, inter: logical ring
-   reduce-scatter, logical ring inter:allgather, intra: binomial broadcast
- - smp_rsag_rab: intra: binomial allreduce, inter: rab
-   reduce-scatter, rab inter:allgather, intra: binomial broadcast
- - redbcast: reduce then broadcast, using default or tuned algorithms if specified
- - ompi_ring_segmented: ring algorithm used by OpenMPI
- - mvapich2_rs: rdb for small messages, reduce-scatter then allgather else
- - mvapich2_two_level: SMP-aware algorithm, with mpich as intra algorithm, and rdb as inter (Change this behavior by using mvapich2 selector to use tuned values)
- - rab: default `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_ implementation
+``default``: naive one, by defautl. |br|
+``ompi``: use openmpi selector for the allreduce operations. |br|
+``mpich``: use mpich selector for the allreduce operations. |br|
+``mvapich2``: use mvapich2 selector for the allreduce operations. |br|
+``impi``: use intel mpi selector for the allreduce operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``lr``: logical ring reduce-scatter then logical ring allgather. |br|
+``rab1``: variations of the  `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_ algorithm: reduce_scatter then allgather. |br|
+``rab2``: variations of the  `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_ algorithm: alltoall then allgather. |br|
+``rab_rsag``: variation of the  `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_ algorithm: recursive doubling reduce_scatter then recursive doubling allgather. |br|
+``rdb``: recursive doubling. |br|
+``smp_binomial``: binomial tree with smp: binomial intra. |br| SMP reduce, inter reduce, inter broadcast then intra broadcast. |br|
+``smp_binomial_pipeline``: same with segment size = 4096 bytes. |br|
+``smp_rdb``: intra``: binomial allreduce, inter: Recursive doubling allreduce, intra``: binomial broadcast. |br|
+``smp_rsag``: intra: binomial allreduce, inter: reduce-scatter, inter:allgather, intra: binomial broadcast. |br|
+``smp_rsag_lr``: intra: binomial allreduce, inter: logical ring reduce-scatter, logical ring inter:allgather, intra: binomial broadcast. |br|
+``smp_rsag_rab``: intra: binomial allreduce, inter: rab reduce-scatter, rab inter:allgather, intra: binomial broadcast. |br|
+``redbcast``: reduce then broadcast, using default or tuned algorithms if specified. |br|
+``ompi_ring_segmented``: ring algorithm used by OpenMPI. |br|
+``mvapich2_rs``: rdb for small messages, reduce-scatter then allgather else. |br|
+``mvapich2_two_level``: SMP-aware algorithm, with mpich as intra algorithm, and rdb as inter (Change this behavior by using mvapich2 selector to use tuned values). |br|
+``rab``: default `Rabenseifner <https://fs.hlrs.de/projects/par/mpi//myreduce.html>`_ implementation. |br|
 
 MPI_Reduce_scatter
 ^^^^^^^^^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the reduce_scatter operations
- - mpich: use mpich selector for the reduce_scatter operations
- - mvapich2: use mvapich2 selector for the reduce_scatter operations
- - impi: use intel mpi selector for the reduce_scatter operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - ompi_basic_recursivehalving: recursive halving version from OpenMPI
- - ompi_ring: ring version from OpenMPI
- - mpich_pair: pairwise exchange version from MPICH
- - mpich_rdb: recursive doubling version from MPICH
- - mpich_noncomm: only works for power of 2 procs, recursive doubling for noncommutative ops
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the reduce_scatter operations. |br|
+``mpich``: use mpich selector for the reduce_scatter operations. |br|
+``mvapich2``: use mvapich2 selector for the reduce_scatter operations. |br|
+``impi``: use intel mpi selector for the reduce_scatter operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``ompi_basic_recursivehalving``: recursive halving version from OpenMPI. |br|
+``ompi_ring``: ring version from OpenMPI. |br|
+``ompi_butterfly``: butterfly version from OpenMPI. |br|
+``mpich_pair``: pairwise exchange version from MPICH. |br|
+``mpich_rdb``: recursive doubling version from MPICH. |br|
+``mpich_noncomm``: only works for power of 2 procs, recursive doubling for noncommutative ops. |br|
 
 
 MPI_Allgather
 ^^^^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the allgather operations
- - mpich: use mpich selector for the allgather operations
- - mvapich2: use mvapich2 selector for the allgather operations
- - impi: use intel mpi selector for the allgather operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - 2dmesh: see alltoall
- - 3dmesh: see alltoall
- - bruck: Described by Bruck et.al. in <a href="http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=642949">
-   Efficient algorithms for all-to-all communications in multiport message-passing systems</a>
- - GB: Gather - Broadcast (uses tuned version if specified)
- - loosely_lr: Logical Ring with grouping by core (hardcoded, default
-   processes/node: 4)
- - NTSLR: Non Topology Specific Logical Ring
- - NTSLR_NB: Non Topology Specific Logical Ring, Non Blocking operations
- - pair: see alltoall
- - rdb: see alltoall
- - rhv: only power of 2 number of processes
- - ring: see alltoall
- - SMP_NTS: gather to root of each SMP, then every root of each SMP node
-   post INTER-SMP Sendrecv, then do INTRA-SMP Bcast for each receiving message,
-   using logical ring algorithm (hardcoded, default processes/SMP: 8)
- - smp_simple: gather to root of each SMP, then every root of each SMP node
-   post INTER-SMP Sendrecv, then do INTRA-SMP Bcast for each receiving message,
-   using simple algorithm (hardcoded, default processes/SMP: 8)
- - spreading_simple: from node i, order of communications is i -> i + 1, i ->
-   i + 2, ..., i -> (i + p -1) % P
- - ompi_neighborexchange: Neighbor Exchange algorithm for allgather.
-   Described by Chen et.al. in  `Performance Evaluation of Allgather
-   Algorithms on Terascale Linux Cluster with Fast Ethernet <http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=1592302>`_
- - mvapich2_smp: SMP aware algorithm, performing intra-node gather, inter-node allgather with one process/node, and bcast intra-node
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the allgather operations. |br|
+``mpich``: use mpich selector for the allgather operations. |br|
+``mvapich2``: use mvapich2 selector for the allgather operations. |br|
+``impi``: use intel mpi selector for the allgather operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``2dmesh``: see alltoall. |br|
+``3dmesh``: see alltoall. |br|
+``bruck``: Described by Bruck et.al. in <a href="http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=642949"> Efficient algorithms for all-to-all communications in multiport message-passing systems</a>. |br|
+``GB``: Gather - Broadcast (uses tuned version if specified). |br|
+``loosely_lr``: Logical Ring with grouping by core (hardcoded, default processes/node: 4). |br|
+``NTSLR``: Non Topology Specific Logical Ring. |br|
+``NTSLR_NB``: Non Topology Specific Logical Ring, Non Blocking operations. |br|
+``pair``: see alltoall. |br|
+``rdb``: see alltoall. |br|
+``rhv``: only power of 2 number of processes. |br|
+``ring``: see alltoall. |br|
+``SMP_NTS``: gather to root of each SMP, then every root of each SMP node. post INTER-SMP Sendrecv, then do INTRA-SMP Bcast for each receiving message, using logical ring algorithm (hardcoded, default processes/SMP: 8). |br|
+``smp_simple``: gather to root of each SMP, then every root of each SMP node post INTER-SMP Sendrecv, then do INTRA-SMP Bcast for each receiving message, using simple algorithm (hardcoded, default processes/SMP: 8). |br|
+``spreading_simple``: from node i, order of communications is i -> i + 1, i -> i + 2, ..., i -> (i + p -1) % P. |br|
+``ompi_neighborexchange``: Neighbor Exchange algorithm for allgather. Described by Chen et.al. in  `Performance Evaluation of Allgather Algorithms on Terascale Linux Cluster with Fast Ethernet <http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=1592302>`_. |br|
+``mvapich2_smp``: SMP aware algorithm, performing intra-node gather, inter-node allgather with one process/node, and bcast intra-node
 
 MPI_Allgatherv
 ^^^^^^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the allgatherv operations
- - mpich: use mpich selector for the allgatherv operations
- - mvapich2: use mvapich2 selector for the allgatherv operations
- - impi: use intel mpi selector for the allgatherv operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - GB: Gatherv - Broadcast (uses tuned version if specified, but only for Bcast, gatherv is not tuned)
- - pair: see alltoall
- - ring: see alltoall
- - ompi_neighborexchange: see allgather
- - ompi_bruck: see allgather
- - mpich_rdb: recursive doubling algorithm from MPICH
- - mpich_ring: ring algorithm from MPICh - performs differently from the  one from STAR-MPI
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the allgatherv operations. |br|
+``mpich``: use mpich selector for the allgatherv operations. |br|
+``mvapich2``: use mvapich2 selector for the allgatherv operations. |br|
+``impi``: use intel mpi selector for the allgatherv operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``GB``: Gatherv - Broadcast (uses tuned version if specified, but only for Bcast, gatherv is not tuned). |br|
+``pair``: see alltoall. |br|
+``ring``: see alltoall. |br|
+``ompi_neighborexchange``: see allgather. |br|
+``ompi_bruck``: see allgather. |br|
+``mpich_rdb``: recursive doubling algorithm from MPICH. |br|
+``mpich_ring``: ring algorithm from MPICh - performs differently from the  one from STAR-MPI.
 
 MPI_Bcast
 ^^^^^^^^^
 
- - default: naive one, by default
- - ompi: use openmpi selector for the bcast operations
- - mpich: use mpich selector for the bcast operations
- - mvapich2: use mvapich2 selector for the bcast operations
- - impi: use intel mpi selector for the bcast operations
- - automatic (experimental): use an automatic self-benchmarking algorithm
- - arrival_pattern_aware: root exchanges with the first process to arrive
- - arrival_pattern_aware_wait: same with slight variation
- - binomial_tree: binomial tree exchange
- - flattree: flat tree exchange
- - flattree_pipeline: flat tree exchange, message split into 8192 bytes pieces
- - NTSB: Non-topology-specific pipelined binary tree with 8192 bytes pieces
- - NTSL: Non-topology-specific pipelined linear with 8192 bytes pieces
- - NTSL_Isend: Non-topology-specific pipelined linear with 8192 bytes pieces, asynchronous communications
- - scatter_LR_allgather: scatter followed by logical ring allgather
- - scatter_rdb_allgather: scatter followed by recursive doubling allgather
- - arrival_scatter: arrival pattern aware scatter-allgather
- - SMP_binary: binary tree algorithm with 8 cores/SMP
- - SMP_binomial: binomial tree algorithm with 8 cores/SMP
- - SMP_linear: linear algorithm with 8 cores/SMP
- - ompi_split_bintree: binary tree algorithm from OpenMPI, with message split in 8192 bytes pieces
- - ompi_pipeline: pipeline algorithm from OpenMPI, with message split in 128KB pieces
- - mvapich2_inter_node: Inter node default mvapich worker
- - mvapich2_intra_node: Intra node default mvapich worker
- - mvapich2_knomial_intra_node:  k-nomial intra node default mvapich worker. default factor is 4.
+``default``: naive one, by default. |br|
+``ompi``: use openmpi selector for the bcast operations. |br|
+``mpich``: use mpich selector for the bcast operations. |br|
+``mvapich2``: use mvapich2 selector for the bcast operations. |br|
+``impi``: use intel mpi selector for the bcast operations. |br|
+``automatic (experimental)``: use an automatic self-benchmarking algorithm. |br|
+``arrival_pattern_aware``: root exchanges with the first process to arrive. |br|
+``arrival_pattern_aware_wait``: same with slight variation. |br|
+``binomial_tree``: binomial tree exchange. |br|
+``flattree``: flat tree exchange. |br|
+``flattree_pipeline``: flat tree exchange, message split into 8192 bytes pieces. |br|
+``NTSB``: Non-topology-specific pipelined binary tree with 8192 bytes pieces. |br|
+``NTSL``: Non-topology-specific pipelined linear with 8192 bytes pieces. |br|
+``NTSL_Isend``: Non-topology-specific pipelined linear with 8192 bytes pieces, asynchronous communications. |br|
+``scatter_LR_allgather``: scatter followed by logical ring allgather. |br|
+``scatter_rdb_allgather``: scatter followed by recursive doubling allgather. |br|
+``arrival_scatter``: arrival pattern aware scatter-allgather. |br|
+``SMP_binary``: binary tree algorithm with 8 cores/SMP. |br|
+``SMP_binomial``: binomial tree algorithm with 8 cores/SMP. |br|
+``SMP_linear``: linear algorithm with 8 cores/SMP. |br|
+``ompi_split_bintree``: binary tree algorithm from OpenMPI, with message split in 8192 bytes pieces. |br|
+``ompi_pipeline``: pipeline algorithm from OpenMPI, with message split in 128KB pieces. |br|
+``mvapich2_inter_node``: Inter node default mvapich worker. |br|
+``mvapich2_intra_node``: Intra node default mvapich worker. |br|
+``mvapich2_knomial_intra_node``:  k-nomial intra node default mvapich worker. default factor is 4.
 
 Automatic Evaluation
 ^^^^^^^^^^^^^^^^^^^^
@@ -447,12 +565,13 @@ Adding an algorithm
 ^^^^^^^^^^^^^^^^^^^
 
 To add a new algorithm, one should check in the src/smpi/colls folder
-how other algorithms are coded. Using plain MPI code inside Simgrid
+how other algorithms are coded. Using plain MPI code inside SimGrid
 can't be done, so algorithms have to be changed to use smpi version of
 the calls instead (MPI_Send will become smpi_mpi_send). Some functions
 may have different signatures than their MPI counterpart, please check
 the other algorithms or contact us using the `>SimGrid
-developers mailing list <http://lists.gforge.inria.fr/mailman/listinfo/simgrid-devel>`_.
+user mailing list <https://sympa.inria.fr/sympa/info/simgrid-community>`_,
+or on `>Mattermost <https://framateam.org/simgrid/channels/town-square>`_.
 
 Example: adding a "pair" version of the Alltoall collective.
 
@@ -464,7 +583,7 @@ Example: adding a "pair" version of the Alltoall collective.
 
  - To register the new version of the algorithm, simply add a line to the corresponding macro in src/smpi/colls/cools.h ( add a "COLL_APPLY(action, COLL_ALLTOALL_SIG, pair)" to the COLL_ALLTOALLS macro ). The algorithm should now be compiled and be selected when using --cfg=smpi/alltoall:pair at runtime.
 
- - To add a test for the algorithm inside Simgrid's test suite, juste add the new algorithm name in the ALLTOALL_COLL list found inside teshsuite/smpi/CMakeLists.txt . When running ctest, a test for the new algorithm should be generated and executed. If it does not pass, please check your code or contact us.
+ - To add a test for the algorithm inside SimGrid's test suite, juste add the new algorithm name in the ALLTOALL_COLL list found inside teshsuite/smpi/CMakeLists.txt . When running ctest, a test for the new algorithm should be generated and executed. If it does not pass, please check your code or contact us.
 
  - Please submit your patch for inclusion in SMPI, for example through a pull request on GitHub or directly per email.
 
@@ -491,124 +610,36 @@ Alltoall on 16 Nodes with the Ring Algorithm.
 
 Alltoall on 16 Nodes with the Pairwise Algorithm.
 
--------------------------
-What can run within SMPI?
--------------------------
+.. _SMPI_mix_s4u:
 
-You can run unmodified MPI applications (both C/C++ and Fortran) within
-SMPI, provided that you only use MPI calls that we implemented. Global
-variables should be handled correctly on Linux systems.
+.............................
+Mixing S4U and MPI simulation
+.............................
 
-....................
-MPI coverage of SMPI
-....................
+Mixing both interfaces is very easy. This can be useful to easily implement a service in S4U that is provided by your
+infrastructure in some way, and test how your MPI application interacts with this service. Or you can use it to start more than
+one MPI application in your simulation, and study their interactions.
 
-Our coverage of the interface is very decent, but still incomplete;
-Given the size of the MPI standard, we may well never manage to
-implement absolutely all existing primitives. Currently, we have
-almost no support for I/O primitives, but we still pass a very large
-amount of the MPICH coverage tests.
+.. doxygenfunction:: SMPI_app_instance_start
 
-The full list of not yet implemented functions is documented in the
-file `include/smpi/smpi.h
-<https://framagit.org/simgrid/simgrid/tree/master/include/smpi/smpi.h>`_
-in your version of SimGrid, between two lines containing the ``FIXME``
-marker. If you really miss a feature, please get in touch with us: we
-can guide you though the SimGrid code to help you implementing it, and
-we'd be glad to integrate your contribution to the main project.
+.. tabs::
 
-.. _SMPI_what_globals:
+   .. group-tab:: Example
 
-.................................
-Privatization of global variables
-.................................
+      Here is a simple example of use, which starts the function ``alltoall_mpi`` as a MPI instance on 4 hosts, along several
+      S4U actors doing a master/workers.
 
-Concerning the globals, the problem comes from the fact that usually,
-MPI processes run as real UNIX processes while they are all folded
-into threads of a unique system process in SMPI. Global variables are
-usually private to each MPI process while they become shared between
-the processes in SMPI.  The problem and some potential solutions are
-discussed in this article: `Automatic Handling of Global Variables for
-Multi-threaded MPI Programs
-<http://charm.cs.illinois.edu/newPapers/11-23/paper.pdf>` (note that
-this article does not deal with SMPI but with a competing solution
-called AMPI that suffers of the same issue).  This point used to be
-problematic in SimGrid, but the problem should now be handled
-automatically on Linux.
+      .. showfile:: examples/smpi/smpi_s4u_masterworker/masterworker_mailbox_smpi.cpp
+         :language: cpp
 
-Older versions of SimGrid came with a script that automatically
-privatized the globals through static analysis of the source code. But
-our implementation was not robust enough to be used in production, so
-it was removed at some point. Currently, SMPI comes with two
-privatization mechanisms that you can :ref:`select at runtime
-<cfg=smpi/privatization>`.  The dlopen approach is used by
-default as it is much faster and still very robust.  The mmap approach
-is an older approach that proves to be slower.
+   .. group-tab:: Deployment file
 
-With the **mmap approach**, SMPI duplicates and dynamically switch the
-``.data`` and ``.bss`` segments of the ELF process when switching the
-MPI ranks. This allows each ranks to have its own copy of the global
-variables.  No copy actually occurs as this mechanism uses ``mmap()``
-for efficiency. This mechanism is considered to be very robust on all
-systems supporting ``mmap()`` (Linux and most BSDs). Its performance
-is questionable since each context switch between MPI ranks induces
-several syscalls to change the ``mmap`` that redirects the ``.data``
-and ``.bss`` segments to the copies of the new rank. The code will
-also be copied several times in memory, inducing a slight increase of
-memory occupation.
-
-Another limitation is that SMPI only accounts for global variables
-defined in the executable. If the processes use external global
-variables from dynamic libraries, they won't be switched
-correctly. The easiest way to solve this is to statically link against
-the library with these globals. This way, each MPI rank will get its
-own copy of these libraries. Of course you should never statically
-link against the SimGrid library itself.
-
-With the **dlopen approach**, SMPI loads several copies of the same
-executable in memory as if it were a library, so that the global
-variables get naturally duplicated. It first requires the executable
-to be compiled as a relocatable binary, which is less common for
-programs than for libraries. But most distributions are now compiled
-this way for security reason as it allows one to randomize the address
-space layout. It should thus be safe to compile most (any?) program
-this way.  The second trick is that the dynamic linker refuses to link
-the exact same file several times, be it a library or a relocatable
-executable. It makes perfectly sense in the general case, but we need
-to circumvent this rule of thumb in our case. To that extend, the
-binary is copied in a temporary file before being re-linked against.
-``dlmopen()`` cannot be used as it only allows 256 contextes, and as it
-would also duplicate simgrid itself.
+      .. showfile:: examples/smpi/smpi_s4u_masterworker/deployment_masterworker_mailbox_smpi.xml
+         :language: xml
 
-This approach greatly speeds up the context switching, down to about
-40 CPU cycles with our raw contextes, instead of requesting several
-syscalls with the ``mmap()`` approach. Another advantage is that it
-permits one to run the SMPI contexts in parallel, which is obviously not
-possible with the ``mmap()`` approach. It was tricky to implement, but
-we are not aware of any flaws, so smpirun activates it by default.
-
-In the future, it may be possible to further reduce the memory and
-disk consumption. It seems that we could `punch holes
-<https://lwn.net/Articles/415889/>`_ in the files before dl-loading
-them to remove the code and constants, and mmap these area onto a
-unique copy. If done correctly, this would reduce the disk- and
-memory- usage to the bare minimum, and would also reduce the pressure
-on the CPU instruction cache. See the `relevant bug
-<https://github.com/simgrid/simgrid/issues/137>`_ on github for
-implementation leads.\n
-
-Also, currently, only the binary is copied and dlopen-ed for each MPI
-rank. We could probably extend this to external dependencies, but for
-now, any external dependencies must be statically linked into your
-application. As usual, simgrid itself shall never be statically linked
-in your app. You don't want to give a copy of SimGrid to each MPI rank:
-that's ways too much for them to deal with.
-
-.. todo: speak of smpi/privatize-libs here
-
-----------------------------------------------
+..............................................
 Adapting your MPI code for further scalability
-----------------------------------------------
+..............................................
 
 As detailed in the `reference article
 <http://hal.inria.fr/hal-01415484>`_, you may want to adapt your code
@@ -617,9 +648,8 @@ hinder the result quality (or even prevent the app to run) if used
 wrongly. We assume that if you want to simulate an HPC application,
 you know what you are doing. Don't prove us wrong!
 
-..............................
 Reducing your memory footprint
-..............................
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 If you get short on memory (the whole app is executed on a single node when
 simulated), you should have a look at the SMPI_SHARED_MALLOC and
@@ -645,9 +675,10 @@ SMPI_SHARED_MALLOC macro in this case, sorry.
 This feature is demoed by the example file
 `examples/smpi/NAS/dt.c <https://framagit.org/simgrid/simgrid/tree/master/examples/smpi/NAS/dt.c>`_
 
-.........................
+.. _SMPI_use_faster:
+
 Toward Faster Simulations
-.........................
+^^^^^^^^^^^^^^^^^^^^^^^^^
 
 If your application is too slow, try using SMPI_SAMPLE_LOCAL,
 SMPI_SAMPLE_GLOBAL and friends to indicate which computation loops can
@@ -656,14 +687,19 @@ their duration, and this duration will be used for the subsequent
 iterations. These samples are done per processor with
 SMPI_SAMPLE_LOCAL, and shared between all processors with
 SMPI_SAMPLE_GLOBAL. Of course, none of this will work if the execution
-time of your loop iteration are not stable.
+time of your loop iteration are not stable. If some parameters have an
+incidence on the timing of a kernel, and if they are reused often
+(same kernel launched with a few different sizes during the run, for example),
+SMPI_SAMPLE_LOCAL_TAG and SMPI_SAMPLE_GLOBAL_TAG can be used, with a tag
+as last parameter, to differentiate between calls. The tag is a character
+chain crafted by the user, with a maximum size of 128, and should include
+what is necessary to group calls of a given size together.
 
 This feature is demoed by the example file
 `examples/smpi/NAS/ep.c <https://framagit.org/simgrid/simgrid/tree/master/examples/smpi/NAS/ep.c>`_
 
-.............................
 Ensuring Accurate Simulations
-.............................
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 Out of the box, SimGrid may give you fairly accurate results, but
 there is a plenty of factors that could go wrong and make your results
@@ -703,23 +739,99 @@ Finally, you may want to check `this article
 <https://hal.inria.fr/hal-00907887>`_ on the classical pitfalls in
 modeling distributed systems.
 
+.. _SMPI_proxy_apps:
+
+......................
+Examples of SMPI Usage
+......................
+
+A small amount of examples can be found directly in the SimGrid
+archive, under `examples/smpi <https://framagit.org/simgrid/simgrid/-/tree/master/examples/smpi>`_.
+Some show how to simply run MPI code in SimGrid, how to use the
+tracing/replay mechanism or how to use plugins written in S4U to
+extend the simulator abilities.
+
+Another source of examples lay in the SimGrid archive, under
+`teshsuite/smpi <https://framagit.org/simgrid/simgrid/-/tree/master/examples/smpi>`_.
+They are not in the ``examples`` directory because they probably don't
+constitute pedagogical examples. Instead, they are intended to stress
+our implementation during the tests. Some of you may be interested
+anyway.
+
+But the best source of SMPI examples is certainly the `proxy app
+<https://framagit.org/simgrid/SMPI-proxy-apps>`_ external project.
+Proxy apps are scale models of real, massive HPC applications: each of
+them exhibits the same communication and computation patterns than the
+massive application that it stands for. But they last only a few
+thousands lines instead of some millions of lines. These proxy apps
+are usually provided for educational purpose, and also to ensure that
+the represented large HPC applications will correctly work with the
+next generation of runtimes and hardware. `This project
+<https://framagit.org/simgrid/SMPI-proxy-apps>`_ gathers proxy apps
+from different sources, along with the patches needed (if any) to run
+them on top of SMPI.
+
+.. _SMPI_offline:
+
+--------------------------
+Offline SMPI: Trace Replay
+--------------------------
+
+Although SMPI is often used for :ref:`online simulation
+<SMPI_online>`, where the application is executed for real, you can
+also go for offline simulation through trace replay.
+
+SimGrid uses time-independent traces, in which each actor is given a
+script of the actions to do sequentially. These trace files can
+actually be captured with the online version of SMPI, as follows:
+
+.. code-block:: console
+
+   $ smpirun -trace-ti --cfg=tracing/filename:LU.A.32 -np 32 -platform ../cluster_backbone.xml bin/lu.A.32
+
+The produced trace is composed of a file ``LU.A.32`` and a folder
+``LU.A.32_files``. The file names don't match with the MPI ranks, but
+that's expected.
+
+To replay this with SMPI, you need to first compile the provided
+``smpi_replay.cpp`` file, that comes from
+`simgrid/examples/smpi/replay
+<https://framagit.org/simgrid/simgrid/tree/master/examples/smpi/replay>`_.
+
+.. code-block:: console
+
+   $ smpicxx ../replay.cpp -O3 -o ../smpi_replay
+
+Afterward, you can replay your trace in SMPI as follows:
+
+.. code-block:: console
+
+   $ smpirun -np 32 -platform ../cluster_torus.xml -ext smpi_replay ../smpi_replay LU.A.32
+
+All the outputs are gone, as the application is not really simulated
+here. Its trace is simply replayed. But if you visualize the live
+simulation and the replay, you will see that the behavior is
+unchanged. The simulation does not run much faster on this very
+example, but this becomes very interesting when your application
+is computationally hungry.
+
 -------------------------
 Troubleshooting with SMPI
 -------------------------
 
-.................................
-./configure refuses to use smpicc
-.................................
+.........................................
+./configure or cmake refuse to use smpicc
+.........................................
 
-If your ``./configure`` reports that the compiler is not
+If your configuration script (such as ``./configure`` or ``cmake``) reports that the compiler is not
 functional or that you are cross-compiling, try to define the
 ``SMPI_PRETEND_CC`` environment variable before running the
 configuration.
 
-.. code-block:: shell
+.. code-block:: console
 
-   SMPI_PRETEND_CC=1 ./configure # here come the configure parameters
-   make
+   $ SMPI_PRETEND_CC=1 ./configure # here come the configure parameters
+   $ make
 
 Indeed, the programs compiled with ``smpicc`` cannot be executed
 without ``smpirun`` (they are shared libraries and do weird things on
@@ -730,21 +842,21 @@ fail without ``smpirun``.
 
 .. warning::
 
-  Make sure that SMPI_PRETEND_CC is only set when calling ./configure,
+  Make sure that SMPI_PRETEND_CC is only set when calling the configuration script but
   not during the actual execution, or any program compiled with smpicc
   will stop before starting.
 
-..............................................
-./configure does not pick smpicc as a compiler
-..............................................
+.....................................................
+./configure or cmake do not pick smpicc as a compiler
+.....................................................
 
 In addition to the previous answers, some projects also need to be
 explicitly told what compiler to use, as follows:
 
-.. code-block:: shell
+.. code-block:: console
 
-   SMPI_PRETEND_CC=1 ./configure CC=smpicc # here come the other configure parameters
-   make
+   $ SMPI_PRETEND_CC=1 cmake CC=smpicc # here come the other configure parameters
+   $ make
 
 Maybe your configure is using another variable, such as ``cc`` (in
 lower case) or similar. Just check the logs.
@@ -753,7 +865,7 @@ lower case) or similar. Just check the logs.
 error: unknown type name 'useconds_t'
 .....................................
 
-Try to add ``-D_GNU_SOURCE`` to your compilation line to get ride
+Try to add ``-D_GNU_SOURCE`` to your compilation line to get rid
 of that error.
 
 The reason is that SMPI provides its own version of ``usleep(3)``
@@ -764,46 +876,6 @@ only if you declare ``_GNU_SOURCE`` before including
 SMPI, then you need to ensure that you pass the right configuration
 defines as advised above.
 
+.. |br| raw:: html
 
-
-.. _SMPI_offline:
-
------------------------------
-Trace Replay and Offline SMPI
------------------------------
-
-Although SMPI is often used for :ref:`online simulation
-<SMPI_online>`, where the application is executed for real, you can
-also go for offline simulation through trace replay.
-
-SimGrid uses time-independent traces, in which each actor is given a
-script of the actions to do sequentially. These trace files can
-actually be captured with the online version of SMPI, as follows:
-
-.. code-block:: shell
-
-   $ smpirun -trace-ti --cfg=tracing/filename:LU.A.32 -np 32 -platform ../cluster_backbone.xml bin/lu.A.32
-
-The produced trace is composed of a file ``LU.A.32`` and a folder
-``LU.A.32_files``. The file names don't match with the MPI ranks, but
-that's expected.
-
-To replay this with SMPI, you need to first compile the provided
-``smpi_replay.cpp`` file, that comes from
-`simgrid/examples/smpi/replay
-<https://framagit.org/simgrid/simgrid/tree/master/examples/smpi/replay>`_.
-
-.. code-block:: shell
-
-   $ smpicxx ../replay.cpp -O3 -o ../smpi_replay
-
-Afterward, you can replay your trace in SMPI as follows:
-
-   $ smpirun -np 32 -platform ../cluster_torus.xml -ext smpi_replay ../smpi_replay LU.A.32
-
-All the outputs are gone, as the application is not really simulated
-here. Its trace is simply replayed. But if you visualize the live
-simulation and the replay, you will see that the behavior is
-unchanged. The simulation does not run much faster on this very
-example, but this becomes very interesting when your application
-is computationally hungry.
+   <br />