Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Proactive Congestion Avoidance for Distributed Deep Learning.

Minkoo Kang1, Gyeongsik Yang1, Yeonho Yoo1

  • 1Department of Computer Science and Engineering, Korea University, 145, Anam-ro, Seongbuk-gu, Seoul 02841, Korea.

Sensors (Basel, Switzerland)
|January 1, 2021
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

Distributed Loads: Problem Solving01:21

Distributed Loads: Problem Solving

922
Beams are structural elements commonly employed in engineering applications requiring different load-carrying capacities. The first step in analyzing a beam under a distributed load is to simplify the problem by dividing the load into smaller regions, which allows one to consider each region separately and calculate the magnitude of the equivalent resultant load acting on each portion of the beam. The magnitude of the equivalent resultant load for each region can be determined by calculating...
922
Parallel Processing01:20

Parallel Processing

447
The brain processes sensory information rapidly due to parallel processing, which involves sending data across multiple neural pathways at the same time. This method allows the brain to manage various sensory qualities, such as shapes, colors, movements, and locations, all concurrently. For instance, when observing a forest landscape, the brain simultaneously processes the movement of leaves, the shapes of trees, the depth between them, and the various shades of green. This enables a quick and...
447
Laminar Flow: Problem Solving01:24

Laminar Flow: Problem Solving

354
Laminar flow occurs when a fluid moves smoothly in parallel layers with minimal mixing and turbulence. In fluid mechanics, ensuring laminar flow within a pipe is essential for precise control of flow characteristics, especially in engineering applications. The key factor in determining whether flow remains laminar is the Reynolds number, a dimensionless quantity that depends on the fluid's velocity, density, viscosity, and the pipe's diameter. A Reynolds number of 2100 or lower...
354
Distributed Loads01:19

Distributed Loads

794
Distributed loads are a common type of load that engineers and scientists encounter in various practical situations. Distributed loads often refer to a type of load spread over a surface or a structure and can be modeled as continuous force per unit area.
For example, consider a bookshelf filled with books stacked vertically adjacent to each other. The weight of the books is evenly distributed over the length of the shelf. As a result, the pressure at different locations on the surface of the...
794
Uniform Depth Channel Flow: Problem Solving01:18

Uniform Depth Channel Flow: Problem Solving

283
To calculate the flow rate for a trapezoidal channel, first, identify the bottom width, side slope, and flow depth of the channel. The cross-sectional area (A) corresponding to the depth of flow (y), channel bottom width (B), and side slope (θ) is determined by:Next, calculate the wetted perimeter, which includes the bottom width and the sloped side lengths in contact with the water. Using the values of the cross-sectional area and the wetted perimeter, determine the hydraulic radius by...
283
Turbulent Flow: Problem Solving01:09

Turbulent Flow: Problem Solving

272
Carbonation is a process used to dissolve carbon dioxide gas in a liquid, commonly used in the production of carbonated beverages. Achieving efficient carbonation requires careful control of temperature, pressure, and flow conditions. By adjusting these parameters, carbonation efficiency can be maximized, producing a higher concentration of CO2 in the liquid.
Temperature is a key factor in CO2 solubility. In this case, the CO2 gas and the liquid are cooled to 20°C. Lower temperatures enhance...
272

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

qCon: QoS-Aware Network Resource Management for Fog Computing.

Sensors (Basel, Switzerland)·2018
Same author

Enhancing the Isolation and Performance of Control Planes for Fog Computing.

Sensors (Basel, Switzerland)·2018
See all related articles

Distributed deep learning (DDL) faces network bottlenecks. Proactive Congestion Notification (PCN) prevents congestion by regulating switch queues, improving DDL traffic throughput by 72% on average.

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Network Engineering

Background:

  • Distributed deep learning (DDL) accelerates model training but suffers from network communication bottlenecks.
  • Synchronizing model gradients during DDL iterations generates burst traffic, leading to network congestion and reduced throughput.

Purpose of the Study:

  • To introduce Proactive Congestion Notification (PCN), a novel congestion-avoidance technique for DDL.
  • To mitigate network congestion caused by DDL's bursty traffic patterns.

Main Methods:

  • PCN proactively regulates switch queue lengths before DDL burst traffic arrives.
  • This regulation prepares network switches to handle the incoming high-volume data efficiently.

Main Results:

Keywords:
P4congestion avoidancedeep learningdistributed deep learningnetwork congestionproactive congestion notification

Related Experiment Videos

  • PCN effectively prevents network congestion in DDL environments.
  • Evaluations demonstrate an average improvement of 72% in DDL traffic throughput.

Conclusions:

  • PCN is a viable solution for enhancing DDL performance.
  • By addressing network bottlenecks, PCN significantly boosts the efficiency of large-scale deep learning training.