Add docs for the distributed namespace (#1184)

Author: angeloskath

docs/src/index.rst

@@ -43,6 +43,7 @@ are the CPU and GPU.
    usage/function_transforms
    usage/compile
    usage/numpy
+   usage/distributed
    usage/using_streams
 
 .. toctree::
@@ -69,6 +70,7 @@ are the CPU and GPU.
    python/metal
    python/nn
    python/optimizers
+   python/distributed
    python/tree_utils
 
 .. toctree::

docs/src/python/distributed.rst

@@ -0,0 +1,19 @@
+.. _distributed:
+
+.. currentmodule:: mlx.core.distributed
+
+Distributed Communication
+==========================
+
+MLX provides a distributed communication package using MPI. The MPI library is
+loaded at runtime; if MPI is available then distributed communication is also
+made available.
+
+.. autosummary::
+   :toctree: _autosummary
+
+    Group
+    is_available
+    init
+    all_sum
+    all_gather

docs/src/usage/distributed.rst

@@ -0,0 +1,166 @@
+.. _usage_distributed:
+
+Distributed Communication
+=========================
+
+.. currentmodule:: mlx.core.distributed
+
+MLX utilizes `MPI <https://en.wikipedia.org/wiki/Message_Passing_Interface>`_ to
+provide distributed communication operations that allow the computational cost
+of training or inference to be shared across many physical machines. You can
+see a list of the supported operations in the :ref:`API docs<distributed>`.
+
+.. note::
+   A lot of operations may not be supported or not as fast as they should be.
+   We are adding more and tuning the ones we have as we are figuring out the
+   best way to do distributed computing on Macs using MLX.
+
+Getting Started
+---------------
+
+MLX already comes with the ability to "talk" to MPI if it is installed on the
+machine. The minimal distributed program in MLX is as simple as:
+
+.. code:: python
+
+    import mlx.core as mx
+
+    world = mx.distributed.init()
+    x = mx.distributed.all_sum(mx.ones(10))
+    print(world.rank(), x)
+
+The program above sums the array ``mx.ones(10)`` across all
+distributed processes. If simply run with ``python``, however, only one
+process is launched and no distributed communication takes place.
+
+To launch the program in distributed mode we need to use ``mpirun`` or
+``mpiexec`` depending on the MPI installation. The simplest possible way is the
+following:
+
+.. code:: shell
+
+    $ mpirun -np 2 python test.py
+    1 array([2, 2, 2, ..., 2, 2, 2], dtype=float32)
+    0 array([2, 2, 2, ..., 2, 2, 2], dtype=float32)
+
+The above launches two processes on the same (local) machine and we can see
+both standard output streams. The processes send the array of 1s to each other
+and compute the sum which is printed. Launching with ``mpirun -np 4 ...`` would
+print 4 etc.
+
+Installing MPI
+---------------
+
+MPI can be installed with Homebrew, using the Anaconda package manager or
+compiled from source. Most of our testing is done using ``openmpi`` installed
+with the Anaconda package manager as follows:
+
+.. code:: shell
+
+    $ conda install openmpi
+
+Installing with Homebrew may require specifying the location of ``libmpi.dyld``
+so that MLX can find it and load it at runtime. This can simply be achieved by
+passing the ``DYLD_LIBRARY_PATH`` environment variable to ``mpirun``.
+
+.. code:: shell
+
+    $ mpirun -np 2 -x DYLD_LIBRARY_PATH=/opt/homebrew/lib/ python test.py
+
+Setting up Remote Hosts
+-----------------------
+
+MPI can automatically connect to remote hosts and set up the communication over
+the network if the remote hosts can be accessed via ssh. A good checklist to
+debug connectivity issues is the following:
+
+* ``ssh hostname`` works from all machines to all machines without asking for
+  password or host confirmation
+* ``mpirun`` is accessible on all machines. You can call ``mpirun`` using its
+  full path to force all machines to use a specific path.
+* Ensure that the ``hostname`` used by MPI is the one that you have configured
+  in the ``.ssh/config`` files on all machines.
+
+.. note::
+  For an example hostname ``foo.bar.com`` MPI can use only ``foo`` as
+  the hostname passed to ssh if the current hostname matches ``*.bar.com``.
+
+An easy way to pass the host names to MPI is using a host file. A host file
+looks like the following, where ``host1`` and ``host2`` should be the fully
+qualified domain names or IPs for these hosts.
+
+.. code::
+
+    host1 slots=1
+    host2 slots=1
+
+When using MLX, it is very likely that you want to use 1 slot per host, ie one
+process per host.  The hostfile also needs to contain the current
+host if you want to run on the local host. Passing the host file to
+``mpirun`` is simply done using the ``--hostfile`` command line argument.
+
+Training Example
+----------------
+
+In this section we will adapt an MLX training loop to support data parallel
+distributed training. Namely, we will average the gradients across a set of
+hosts before applying them to the model.
+
+Our training loop looks like the following code snippet if we omit the model,
+dataset and optimizer initialization.
+
+.. code:: python
+
+    model = ...
+    optimizer = ...
+    dataset = ...
+
+    def step(model, x, y):
+        loss, grads = loss_grad_fn(model, x, y)
+        optimizer.update(model, grads)
+        return loss
+
+    for x, y in dataset:
+        loss = step(model, x, y)
+        mx.eval(loss, model.parameters())
+
+All we have to do to average the gradients across machines is perform an
+:func:`all_sum` and divide by the size of the :class:`Group`. Namely we
+have to :func:`mlx.utils.tree_map` the gradients with following function.
+
+.. code:: python
+
+    def all_avg(x):
+        return mx.distributed.all_sum(x) / mx.distributed.init().size()
+
+Putting everything together our training loop step looks as follows with
+everything else remaining the same.
+
+.. code:: python
+
+    from mlx.utils import tree_map
+
+    def all_reduce_grads(grads):
+        N = mx.distributed.init()
+        if N == 1:
+            return grads
+        return tree_map(
+                lambda x: mx.distributed.all_sum(x) / N,
+                grads)
+
+    def step(model, x, y):
+        loss, grads = loss_grad_fn(model, x, y)
+        grads = all_reduce_grads(grads)  # <--- This line was added
+        optimizer.update(model, grads)
+        return loss
+
+Tuning All Reduce
+-----------------
+
+We are working on improving the performance of all reduce on MLX but for now
+the two main things one can do to extract the most out of distributed training with MLX are:
+
+1. Perform a few large reductions instead of many small ones to improve
+   bandwidth and latency
+2. Pass ``--mca btl_tcp_links 4`` to ``mpirun`` to configure it to use 4 tcp
+   connections between each host to improve bandwidth

examples/cpp/distributed.cpp

@@ -16,7 +16,7 @@ int main() {
   std::cout << global_group.rank() << " / " << global_group.size() << std::endl;
 
   array x = ones({10});
-  array out = distributed::all_reduce_sum(x, global_group);
+  array out = distributed::all_sum(x, global_group);
 
   std::cout << out << std::endl;
 }

mlx/distributed/distributed.h

@@ -56,7 +56,7 @@ namespace detail {
 Stream communication_stream();
 
 /* Perform an all reduce sum operation */
-void all_reduce_sum(Group group, const array& input, array& output);
+void all_sum(Group group, const array& input, array& output);
 
 /* Perform an all reduce sum operation */
 void all_gather(Group group, const array& input, array& output);

mlx/distributed/mpi/mpi.cpp

@@ -260,7 +260,7 @@ Stream communication_stream() {
   return comm_stream;
 }
 
-void all_reduce_sum(Group group, const array& input_, array& output) {
+void all_sum(Group group, const array& input_, array& output) {
   array input = ensure_row_contiguous(input_);
   mpi().all_reduce(
       (input.data<void>() == output.data<void>()) ? MPI_IN_PLACE

mlx/distributed/no_distributed.cpp

@@ -31,7 +31,7 @@ Stream communication_stream() {
   return comm_stream;
 }
 
-void all_reduce_sum(Group group, const array& input, array& output) {}
+void all_sum(Group group, const array& input, array& output) {}
 void all_gather(Group group, const array& input, array& output) {}
 
 } // namespace detail

mlx/distributed/ops.cpp

@@ -17,7 +17,7 @@ Group to_group(std::optional<Group> group) {
 
 } // namespace
 
-array all_reduce_sum(const array& x, std::optional<Group> group_) {
+array all_sum(const array& x, std::optional<Group> group_) {
   auto group = to_group(group_);
 
   if (group.size() == 1) {

mlx/distributed/ops.h

@@ -8,7 +8,7 @@
 
 namespace mlx::core::distributed {
 
-array all_reduce_sum(const array& x, std::optional<Group> group = std::nullopt);
+array all_sum(const array& x, std::optional<Group> group = std::nullopt);
 array all_gather(const array& x, std::optional<Group> group = std::nullopt);
 
 } // namespace mlx::core::distributed

mlx/distributed/primitives.cpp

@@ -24,7 +24,7 @@ void AllReduce::eval_cpu(
 
   switch (reduce_type_) {
     case Sum:
-      distributed::detail::all_reduce_sum(group(), inputs[0], outputs[0]);
+      distributed::detail::all_sum(group(), inputs[0], outputs[0]);
       break;
     default:
       throw std::runtime_error("Only all reduce sum is supported for now");
@@ -36,7 +36,7 @@ std::pair<std::vector<array>, std::vector<int>> AllReduce::vmap(
     const std::vector<int>& axes) {
   switch (reduce_type_) {
     case Sum:
-      return {{all_reduce_sum(inputs[0], group())}, axes};
+      return {{all_sum(inputs[0], group())}, axes};
     default:
       throw std::runtime_error("Only all reduce sum is supported for now");
   }
@@ -48,7 +48,7 @@ std::vector<array> AllReduce::jvp(
     const std::vector<int>& argnums) {
   switch (reduce_type_) {
     case Sum:
-      return {all_reduce_sum(tangents[0], group())};
+      return {all_sum(tangents[0], group())};
     default:
       throw std::runtime_error("Only all reduce sum is supported for now");
   }