HPGO

Hybrid Parallelism Global Orchestration

(Project Name TBD)

Get Started

Install

From Rust Cargo, as a Rust crate

bash cargo install HPGO

From Python PyPI, as a Python3 package

bash pip3 install HPGO

Build from source

```bash rustup default nightly cargo build --release

library produced under targets/release/libHPGO.so

```

Use

```python

Import HPGO Python API from HPGO.so

import HPGO

Construct the Conductor object

c = HPGO.conductorfromtorchgraphandseps("./profiles/xlnet/graph.txt", 64, 512, [8, 16]) res = c.pyorchestrate() print(res) ```

License

This project is open sourced under the terms of BSD-3-Clause, details of which can be found in the LICENSE file

If you apply this library to any project and research, please cite this code:

@misc{HPGO, author = {Yi Rong}, title = {HPGO, [{Hybrid Parallelism} Global Orchestration]}, howpublished = {\url{https://github.com/LER0ever/HPGO}}, year = {2020} }

The project contains source code from PipeDream, a Microsoft Research project licensed under the MIT License. Code taken from Pipedream includes a Rust file src/input/torch_graph_py.rs, several profiling data files under the profiles directory, and the contrib directory.

This project also contains profiling data provided by the AlibabaPAI/DAPPLE project. The development of the DAPPLE planner continues here.