Name | torchmonarch-nightly JSON |
Version |
2025.7.27
JSON |
| download |
home_page | None |
Summary | Monarch: Single controller library |
upload_time | 2025-07-27 00:45:58 |
maintainer | None |
docs_url | None |
author | Meta |
requires_python | >=3.10 |
license | BSD-3-Clause |
keywords |
|
VCS |
|
bugtrack_url |
|
requirements |
No requirements were recorded.
|
Travis-CI |
No Travis.
|
coveralls test coverage |
No coveralls.
|
# Monarch 🦋
**Monarch** is a distributed execution engine for PyTorch. Our overall goal is
to deliver the high-quality user experience that people get from single-GPU
PyTorch, but at cluster scale.
> ⚠️ **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.
Note: Monarch is currently only supported on Linux systems
## Installation
### On Fedora distributions
`pip install torchmonarch-nightly`
or manually
```sh
# 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 envrionments, 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 .
# Run unit tests. consider -s for more verbose output
pytest python/tests/ -v -m "not oss_skip"
```
### 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).
```sh
# 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 .
```
## 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!
## License
Monarch is BSD-3 licensed, as found in the [LICENSE](LICENSE) file.
Raw data
{
"_id": null,
"home_page": null,
"name": "torchmonarch-nightly",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.10",
"maintainer_email": null,
"keywords": null,
"author": "Meta",
"author_email": "oncall+monarch@xmail.facebook.com",
"download_url": null,
"platform": null,
"description": "# Monarch \ud83e\udd8b\n\n**Monarch** is a distributed execution engine for PyTorch. Our overall goal is\nto deliver the high-quality user experience that people get from single-GPU\nPyTorch, but at cluster scale.\n\n> \u26a0\ufe0f **Early Development Warning** Monarch is currently in an experimental\n> stage. You should expect bugs, incomplete features, and APIs that may change\n> in future versions. The project welcomes bugfixes, but to make sure things are\n> well coordinated you should discuss any significant change before starting the\n> work. It's recommended that you signal your intention to contribute in the\n> issue tracker, either by filing a new issue or by claiming an existing one.\n\nNote: Monarch is currently only supported on Linux systems\n\n## Installation\n\n### On Fedora distributions\n\n`pip install torchmonarch-nightly`\n\nor manually\n\n```sh\n\n# Create and activate the conda environment\nconda create -n monarchenv python=3.10 -y\nconda activate monarchenv\n\n# Install nightly rust toolchain\ncurl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh\nrustup toolchain install nightly\nrustup default nightly\n\n# Install non-python dependencies\nconda install libunwind -y\n\n# Install the correct cuda and cuda-toolkit versions for your machine\nsudo dnf install cuda-toolkit-12-0 cuda-12-0\n\n# Install clang-dev and nccl-dev\nsudo dnf install clang-devel libnccl-devel\n# Or, in some envrionments, the following may be necessary instead\nconda install -c conda-forge clangdev nccl\nconda update -n monarchenv --all -c conda-forge -y\n\n# 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\nsudo dnf install -y libibverbs rdma-core libmlx5 libibverbs-devel rdma-core-devel\n\n# Install build dependencies\npip install -r build-requirements.txt\n# Install test dependencies\npip install -r python/tests/requirements.txt\n\n# Build and install Monarch\npip install --no-build-isolation .\n# or setup for development\npip install --no-build-isolation -e .\n\n# Run unit tests. consider -s for more verbose output\npytest python/tests/ -v -m \"not oss_skip\"\n```\n\n### On MacOS\n\nYou can also build Monarch to run locally on a MacOS system.\n\nNote that this does not support tensor engine, which is tied to CUDA and RDMA (via ibverbs).\n\n\n```sh\n\n# Create and activate the conda environment\nconda create -n monarchenv python=3.10 -y\nconda activate monarchenv\n\n# Install nightly rust toolchain\ncurl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh\nrustup toolchain install nightly\nrustup default nightly\n\n# Install build dependencies\npip install -r build-requirements.txt\n# Install test dependencies\npip install -r python/tests/requirements.txt\n\n# Build and install Monarch\nUSE_TENSOR_ENGINE=0 pip install --no-build-isolation .\n# or setup for development\nUSE_TENSOR_ENGINE=0 pip install --no-build-isolation -e .\n\n```\n\n\n## Running examples\n\nCheck out the `examples/` directory for demonstrations of how to use Monarch's APIs.\n\nWe'll be adding more examples as we stabilize and polish functionality!\n\n## License\n\nMonarch is BSD-3 licensed, as found in the [LICENSE](LICENSE) file.\n",
"bugtrack_url": null,
"license": "BSD-3-Clause",
"summary": "Monarch: Single controller library",
"version": "2025.7.27",
"project_urls": null,
"split_keywords": [],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "d2eb2fff7b1fcc488b07ea22510fa3b980a5ff1047520796182770b9b055282a",
"md5": "24479afd3f05cfa58d8e246d2589cc50",
"sha256": "60235670551a9a29f729b8e70ccdf3dc7d27eb50dec9f36e6b52afce962ce851"
},
"downloads": -1,
"filename": "torchmonarch_nightly-2025.7.27-cp310-cp310-manylinux2014_x86_64.whl",
"has_sig": false,
"md5_digest": "24479afd3f05cfa58d8e246d2589cc50",
"packagetype": "bdist_wheel",
"python_version": "cp310",
"requires_python": ">=3.10",
"size": 25147659,
"upload_time": "2025-07-27T00:45:58",
"upload_time_iso_8601": "2025-07-27T00:45:58.080808Z",
"url": "https://files.pythonhosted.org/packages/d2/eb/2fff7b1fcc488b07ea22510fa3b980a5ff1047520796182770b9b055282a/torchmonarch_nightly-2025.7.27-cp310-cp310-manylinux2014_x86_64.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "d86c547c585484535ea660aa8c73f32336a72823b505bc955775a63634a3db11",
"md5": "3413611cf15a1f9a4ae349cc73440da5",
"sha256": "4dc64ac6047f5c120389e35a812ab0739eb176cac2567762bee032b27cd639b2"
},
"downloads": -1,
"filename": "torchmonarch_nightly-2025.7.27-cp311-cp311-manylinux2014_x86_64.whl",
"has_sig": false,
"md5_digest": "3413611cf15a1f9a4ae349cc73440da5",
"packagetype": "bdist_wheel",
"python_version": "cp311",
"requires_python": ">=3.10",
"size": 25156281,
"upload_time": "2025-07-27T00:46:00",
"upload_time_iso_8601": "2025-07-27T00:46:00.957187Z",
"url": "https://files.pythonhosted.org/packages/d8/6c/547c585484535ea660aa8c73f32336a72823b505bc955775a63634a3db11/torchmonarch_nightly-2025.7.27-cp311-cp311-manylinux2014_x86_64.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "0830bdf8206dd50bebc29c48539b38c48600f5df04cdc75a19a32d8a5a4bafbc",
"md5": "7cf6553e826e3247d28171165d037897",
"sha256": "499c7626a1c79a5525ef3731db18dc9fe8149d5483327886621154043a11d3ff"
},
"downloads": -1,
"filename": "torchmonarch_nightly-2025.7.27-cp312-cp312-manylinux2014_x86_64.whl",
"has_sig": false,
"md5_digest": "7cf6553e826e3247d28171165d037897",
"packagetype": "bdist_wheel",
"python_version": "cp312",
"requires_python": ">=3.10",
"size": 25201866,
"upload_time": "2025-07-27T00:46:04",
"upload_time_iso_8601": "2025-07-27T00:46:04.101748Z",
"url": "https://files.pythonhosted.org/packages/08/30/bdf8206dd50bebc29c48539b38c48600f5df04cdc75a19a32d8a5a4bafbc/torchmonarch_nightly-2025.7.27-cp312-cp312-manylinux2014_x86_64.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "e9513e0fb0cbeaae4b8c89302362e20c63fc0ec00bb81fbf09b2d9c5bae5f9c8",
"md5": "e19ae3a9c6a2824c6034a90006ec1e58",
"sha256": "27aedcf918f21f8c986438e61456b595e3b9a39cfcc6a8baae3df9e46ddc20fe"
},
"downloads": -1,
"filename": "torchmonarch_nightly-2025.7.27-cp313-cp313-manylinux2014_x86_64.whl",
"has_sig": false,
"md5_digest": "e19ae3a9c6a2824c6034a90006ec1e58",
"packagetype": "bdist_wheel",
"python_version": "cp313",
"requires_python": ">=3.10",
"size": 25201405,
"upload_time": "2025-07-27T00:46:06",
"upload_time_iso_8601": "2025-07-27T00:46:06.957626Z",
"url": "https://files.pythonhosted.org/packages/e9/51/3e0fb0cbeaae4b8c89302362e20c63fc0ec00bb81fbf09b2d9c5bae5f9c8/torchmonarch_nightly-2025.7.27-cp313-cp313-manylinux2014_x86_64.whl",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-07-27 00:45:58",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "torchmonarch-nightly"
}