Concepedia

Pipeline Partitioning

Pipeline Partitioning is a method in distributed deep learning that assigns segments of a neural network model to distinct devices along a pipeline, where the differentiation lies in its strategic division of model layers into contiguous stages to optimize computational efficiency and minimize inter-device communication overhead.

27

Publications

2.1K

Citations

112

Authors

41

Institutions

Publications per year

2017–2026

24

Parents

Publications

27

PipeDream

Deepak Narayanan, Aaron Harlap, Amar Phanishayee et al. · 2019 · 752 citations

, , +17

MAERI

Hyoukjun Kwon, Ananda Samajdar, Tushar Krishna · 2018 · 280 citations

, , +17

MAERI

Hyoukjun Kwon, Ananda Samajdar, Tushar Krishna · ACM SIGPLAN Notices · 2018 · 203 citations

, , +17

PipeDream: Fast and Efficient Pipeline Parallel DNN Training

Aaron Harlap, Deepak Narayanan, Amar Phanishayee et al. · arXiv (Cornell University) · 2018 · 97 citations · Full text

, , +17

Efficient and Robust Parallel DNN Training through Model Parallelism on Multi-GPU Platform

Chi‐Chung Chen, Chia-Lin Yang, Hsiang-Yun Cheng · arXiv (Cornell University) · 2018 · 78 citations · Full text

, , +17

1–5 of 27

Authors

112

  1. AP
    Amar Phanishayee

    Microsoft Research (United Kingdom), United Kingdom

    3 publications · 864 citations

    3

    h-index

  2. NR
    Nikhil R. Devanur

    Microsoft (United States), United States

    3 publications · 864 citations

    3

    h-index

  3. GR
    Gregory R. Ganger

    Carnegie Mellon University, United States

    2 publications · 849 citations

    2

    h-index

  4. AS
    Ananda Samajdar

    Georgia Institute of Technology, United States

    2 publications · 483 citations

    2

    h-index

  5. SL
    Shengwei Li

    National University of Defense Technology, China

    2 publications · 82 citations

    2

    h-index

1–5 of 112

Institutions

41

  1. Georgia Institute of Technology

    Atlanta, United States

    3 h-index

    10 publications

    1.5K citations

  2. University of Science and Technology of China

    Hefei, China

    2 h-index

    3 publications

    84 citations

  3. University of Southern California

    Los Angeles, United States

    2 h-index

    5 publications

    330 citations

  4. National University of Defense Technology

    Changsha, China

    2 h-index

    14 publications

    612 citations

  5. University of Hong Kong

    Pok Fu Lam, Hong Kong

    2 h-index

    12 publications

    468 citations

  6. Twin Cities Orthopedics

    Edina, United States

    1 h-index

    1 publications

    21 citations

  7. Xilinx (Ireland)

    Dublin, Ireland

    1 h-index

    4 publications

    176 citations

  8. École Polytechnique Fédérale de Lausanne

    Lausanne, Switzerland

    1 h-index

    1 publications

    15 citations

1–8 of 41

Venues

  1. IEEE Transactions on Mobile Computing

    TMCCI · 3 publications · 113 citations

    3

    h-index

  2. arXiv (Cornell University)

    Repository · 3 publications · 190 citations

    3

    h-index

  3. IEEE Transactions on Parallel and Distributed Systems

    TPDS · 2 publications · 106 citations

    2

    h-index

  4. 1

    h-index

  5. ACM SIGPLAN Notices

    1 publications · 203 citations

    1

    h-index

1–5