torchmonarch 0.1.0rc8

Monarch: Single controller library

Monarch 🦋

Monarch is a distributed programming framework for PyTorch based on scalable actor messaging. It provides:

1. Remote actors with scalable messaging: Actors are grouped into collections called meshes and messages can be broadcast to all members.
2. Fault tolerance through supervision trees: Actors and processes form a tree and failures propagate up the tree, providing good default error behavior and enabling fine-grained fault recovery.
3. Point-to-point RDMA transfers: cheap registration of any GPU or CPU memory in a process, with the one-sided transfers based on libibverbs
4. Distributed tensors: actors can work with tensor objects sharded across processes

Monarch code imperatively describes how to create processes and actors using a simple python API:

from monarch.actor import Actor, endpoint, this_host

# spawn 8 trainer processes one for each gpu
training_procs = this_host().spawn_procs({"gpus": 8})


# define the actor to run on each process
class Trainer(Actor):
    @endpoint
    def train(self, step: int): ...


# create the trainers
trainers = training_procs.spawn("trainers", Trainer)

# tell all the trainers to take a step
fut = trainers.train.call(step=0)

# wait for all trainers to complete
fut.get()

The introduction to monarch concepts provides an introduction to using these features.

⚠️ Early Development Warning Monarch is currently in an experimental stage. You should expect bugs, incomplete features, and APIs that may change in future versions. The project welcomes bugfixes, but to make sure things are well coordinated you should discuss any significant change before starting the work. It's recommended that you signal your intention to contribute in the issue tracker, either by filing a new issue or by claiming an existing one.

📖 Documentation

View Monarch's hosted documentation at this link.

Installation

Note for running distributed tensors, the local torch version must match the version that monarch was built with.

Stable

pip install torchmonarch

torchmonarch stable is built with the latest stable torch.

Nightly

pip install torchmonarch-nightly

torchmonarch-nightly is built with torch nightly.

Build and Install from Source

On Fedora distributions

# Create and activate the conda environment
conda create -n monarchenv python=3.10 -y
conda activate monarchenv

# Install nightly rust toolchain
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
rustup toolchain install nightly
rustup default nightly

# Install non-python dependencies
conda install libunwind -y

# Install the correct cuda and cuda-toolkit versions for your machine
sudo dnf install cuda-toolkit-12-0 cuda-12-0

# Install clang-dev and nccl-dev
sudo dnf install clang-devel libnccl-devel
# Or, in some environments, the following may be necessary instead
conda install -c conda-forge clangdev nccl
conda update -n monarchenv --all -c conda-forge -y

# If you are building with RDMA support, build monarch with `USE_TENSOR_ENGINE=1 pip install --no-build-isolation .` and dnf install the following packages
sudo dnf install -y libibverbs rdma-core libmlx5 libibverbs-devel rdma-core-devel

# Install build dependencies
pip install -r build-requirements.txt
# Install test dependencies
pip install -r python/tests/requirements.txt

# Build and install Monarch
pip install --no-build-isolation .
# or setup for development
pip install --no-build-isolation -e .

# Verify installation
pip list | grep monarch

On Ubuntu distributions

# Clone the repository and navigate to it
git clone https://github.com/meta-pytorch/monarch.git
cd monarch

# Install nightly rust toolchain
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
source $HOME/.cargo/env
rustup toolchain install nightly
rustup default nightly

# Install Ubuntu-specific system dependencies
sudo apt install -y ninja-build
sudo apt install -y libunwind-dev
sudo apt install -y clang

# Set clang as the default C/C++ compiler
export CC=clang
export CXX=clang++

# Install build dependencies
pip install -r build-requirements.txt
# Install test dependencies
pip install -r python/tests/requirements.txt

# Build and install Monarch (with tensor engine support)
pip install --no-build-isolation .

# or
# Build and install Monarch (without tensor engine support)
USE_TENSOR_ENGINE=0 pip install --no-build-isolation .

# or setup for development
pip install --no-build-isolation -e .

# Verify installation
pip list | grep monarch

On MacOS

You can also build Monarch to run locally on a MacOS system.

Note that this does not support tensor engine, which is tied to CUDA and RDMA (via ibverbs).

# Create and activate the conda environment
conda create -n monarchenv python=3.10 -y
conda activate monarchenv

# Install nightly rust toolchain
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
rustup toolchain install nightly
rustup default nightly

# Install build dependencies
pip install -r build-requirements.txt
# Install test dependencies
pip install -r python/tests/requirements.txt

# Build and install Monarch
USE_TENSOR_ENGINE=0 pip install --no-build-isolation .
# or setup for development
USE_TENSOR_ENGINE=0 pip install --no-build-isolation -e .

# Verify installation
pip list | grep monarch

Running examples

Check out the examples/ directory for demonstrations of how to use Monarch's APIs.

We'll be adding more examples as we stabilize and polish functionality!

Running tests

We have both Rust and Python unit tests. Rust tests are run with cargo-nextest and Python tests are run with pytest.

Rust tests:

# We use cargo-nextest to run our tests, as they can provide strong process isolation
# between every test.
# Here we install it from source, but you can instead use a pre-built binary described
# here: https://nexte.st/docs/installation/pre-built-binaries/
cargo install cargo-nextest --locked
cargo nextest run

cargo-nextest supports all of the filtering flags of "cargo test".

Python tests:

# Make sure to install test dependencies first
pip install -r python/tests/requirements.txt
# Run unit tests. consider -s for more verbose output
pytest python/tests/ -v -m "not oss_skip"

License

Monarch is BSD-3 licensed, as found in the LICENSE file.