GithubHelp home page GithubHelp logo

joeyyoung / mlcc Goto Github PK

View Code? Open in Web Editor NEW
1.0 1.0 0.0 5.55 MB

a distributed flow control protocol for ML cluster

License: MIT License

Rust 14.96% Python 37.91% C++ 6.55% Makefile 22.74% Shell 0.90% C 16.53% LLVM 0.42%

mlcc's Introduction

MLCC

Codes for "Dynamic Flow Scheduling for DNN Training Workloads in Data Centers".

Install

Softwares

Dependency Version
OS Ubuntu-18.04
OS Kernel Linux 5.4.0-77-generic
GCC gcc 7.5.0
CUDA-Toolkit cuda 10.2
OpenMPI openmpi 4.1.1
Horovod v0.22.0
BytePS v0.2.4
Python python3.6
Rust rustc 1.58.0-nightly
NCCL Custormized based on v2.11.4

The software dependency is listed in the table above. All can be downloaded from official repositories.

See ./proto/nccl_patch for the customizations of nccl.

Testbed

6 servers each with two GTX 1080Ti GPUs.

Launch datapath program in kernel space

cd ./proto/kernel && make && sudo ./ccp_kernel_load ipc=0

use sh ./proto/rebuild_ccp.sh to rebuild if the datapath program changes.

Start

On a dedicated server, use switch.py to start an emulated switch with rpc services.

On each server, run distributed agent with ccp.py.

Inject workloads with template scripts in ./proto/workloads.

See ./bbr ./reno-cubic ./deepcc for part of baseline implementations.

mlcc's People

Contributors

joeyyoung avatar

Stargazers

avatar

Watchers

avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.