All Tags
AWS
algorithm-design
architecture
cloud-principles
cost-reduction
data-centric
data-compression
data-processing
deployment
design
edge-computing
energy-footprint
hardware
libraries
locality
machine-learning
management
measured
migration
model-optimization
model-training
performance
queries
rebuilding
scaling
services
strategies
template
workloads
Tactic: Use Power Capping
Tactic sort:
Awesome Tactic
Type: Architectural Tactic
Category: green-ml-enabled-systems
Title
Use Power Capping
Description
Power capping is a technique used to limit the amount of power consumed by a device or system, such as a CPU, GPU, or server. It involves setting a maximum power consumption threshold for the device, and dynamically adjusting the power usage to ensure that it stays below that threshold. This is typically done to manage the power consumption and heat dissipa- tion of a device, and to prevent it from exceeding the power budget of a data center or other power-limited environment.
Participant
Software Designer
Related software artifact
Hardware
Context
Machine Learning
Software feature
< unknown >
Tactic intent
Improve energy efficiency by using power capping to limit the energy usage of a ML model
Target quality attribute
Energy Efficiency
Other related quality attributes
Performance
Measured impact
Krzywaniak et al show that restricting the use of GPU resources can lead to reduced performance and longer execution times, but in certain configurations, it can also result in a significant reduction in energy consumption (up to 33%) with a moderate impact on performance.
Source
Adam Krzywaniak, Pawel Czarnul, and Jerzy Proficz. 2022. GPU Power Capping for Energy-Performance Trade-Offs in Training of Deep Convolutional Neural Networks for Image Recognition. In International Conference on Computational Science. Springer, 667–681. (DOI: https://doi.org/10.1007/978-3-031-08751-6_48)Graphical representation
