Power Modeling and Power Management Framework Dexin Li December 2002.

Slides:



Advertisements
Similar presentations
Embedded System, A Brief Introduction
Advertisements

System Integration and Performance
Seyedehmehrnaz Mireslami, Mohammad Moshirpour, Behrouz H. Far Department of Electrical and Computer Engineering University of Calgary, Canada {smiresla,
The System and Software Development Process Instructor: Dr. Hany H. Ammar Dept. of Computer Science and Electrical Engineering, WVU.
Jill Goryca, Richard Hill American Control Conference June 17, 2013.
DISTRIBUTED CONSISTENCY MANAGEMENT IN A SINGLE ADDRESS SPACE DISTRIBUTED OPERATING SYSTEM Sombrero.
Analysis of power dissipation in embedded systems using real-time operating systems Dick, R.P. Lakshminarayana, G. Raghunathan, A. Jha, N.K. Dept. of Electr.
Mahapatra-Texas A&M-Fall'001 cosynthesis Introduction to cosynthesis Rabi Mahapatra CPSC498.
Chapter 13 Embedded Systems
Scheduling with Optimized Communication for Time-Triggered Embedded Systems Slide 1 Scheduling with Optimized Communication for Time-Triggered Embedded.
Memory Management, File Systems, I/O How Multiprogramming Issues Mesh ECEN 5043 Software Engineering of Multiprogram Systems University of Colorado, Boulder.
2008EECS Embedded Network Programming nesC, TinyOS, Networking, Microcontrollers Jonathan Hui University of California, Berkeley.
DIVES Alur, Lee, Kumar, Pappas: University of Pennsylvania  Charon: high-level modeling language and a design environment reflecting the current state.
Mehmet Can Vuran, Instructor University of Nebraska-Lincoln Acknowledgement: Overheads adapted from those provided by the authors of the textbook.
System-Level Types for Component-Based Design Paper by: Edward A. Lee and Yuhong Xiong Presentation by: Dan Patterson.
BRASS Analysis of QuasiStatic Scheduling Techniques in a Virtualized Reconfigurable Machine Yury Markovskiy, Eylon Caspi, Randy Huang, Joseph Yeh, Michael.
Slide 3-1 Copyright © 2004 Pearson Education, Inc. Operating Systems: A Modern Perspective, Chapter 3 Operating System Organization.
An energy-aware framework for dynamic software management in mobile computing systems Yunsi Fei, Lin Zhong, and Niraj K. Jha Presented By Vimarsh Puneet.
Copyright Arshi Khan1 System Programming Instructor Arshi Khan.
Power is Leading Design Constraint Direct Impacts of Power Management – IDC: Server 2% of US energy consumption and growing exponentially HPC cluster market.
Chapter 10 Architectural Design
4.x Performance Technology drivers – Exascale systems will consist of complex configurations with a huge number of potentially heterogeneous components.
Operating Systems CS3502 Fall 2014 Dr. Jose M. Garrido
Contact Information Office: 225 Neville Hall Office Hours: Monday and Wednesday 12:00-1:00 and by appointment.
CHAPTER 3 TOP LEVEL VIEW OF COMPUTER FUNCTION AND INTERCONNECTION
SOFTWARE DESIGN (SWD) Instructor: Dr. Hany H. Ammar
1 Advanced topics in OpenCIM 1.CIM: The need and the solution.CIM: The need and the solution. 2.Architecture overview.Architecture overview. 3.How Open.
Network-on-Chip Energy-Efficient Design Techniques for Interconnects Suhail Basit.
2009 Sep 10SYSC Dept. Systems and Computer Engineering, Carleton University F09. SYSC2001-Ch7.ppt 1 Chapter 7 Input/Output 7.1 External Devices 7.2.
Mobile Middleware for Energy-Awareness Wei Li
System Architecture Directions for Networked Sensors Jason Hill, Robert Szewczyk, Alec Woo, Seth Hollar, David Culler, Kris Pister Presented by Yang Zhao.
1 The Five Parts of an Information System
System Architecture of Sensor Network Processors Alan Pilecki.
CE Operating Systems Lecture 3 Overview of OS functions and structure.
Slide 3-1 Copyright © 2004 Pearson Education, Inc. Operating Systems: A Modern Perspective, Chapter 3.
The System and Software Development Process Instructor: Dr. Hany H. Ammar Dept. of Computer Science and Electrical Engineering, WVU.
Distributed System Concepts and Architectures 2.3 Services Fall 2011 Student: Fan Bai
INFORMATION SYSTEM-SOFTWARE Topic: OPERATING SYSTEM CONCEPTS.
Distribution and components. 2 What is the problem? Enterprise computing is Large scale & complex: It supports large scale and complex organisations Spanning.
Next Generation Operating Systems Zeljko Susnjar, Cisco CTG June 2015.
Service-oriented Resource Broker for QoS-Guaranteed in Grid Computing System Yichao Yang, Jin Wu, Lei Lang, Yanbo Zhou and Zhili Sun Centre for communication.
CMPS 435 F08 These slides are designed to accompany Web Engineering: A Practitioner’s Approach (McGraw-Hill 2008) by Roger Pressman and David Lowe, copyright.
Abstract A Structured Approach for Modular Design: A Plug and Play Middleware for Sensory Modules, Actuation Platforms, Task Descriptions and Implementations.
Design and Implementation of Spacecraft Avionics Software Architecture based on Spacecraft Onboard Interface Services and Packet Utilization Standard Beijing.
SOFTWARE DESIGN AND ARCHITECTURE LECTURE 31. Review Creational Design Patterns – Singleton Pattern – Builder Pattern.
Data Center & Large-Scale Systems (updated) Luis Ceze, Bill Feiereisen, Krishna Kant, Richard Murphy, Onur Mutlu, Anand Sivasubramanian, Christos Kozyrakis.
Course: COMS-E6125 Professor: Gail E. Kaiser Student: Shanghao Li (sl2967)
Software Engineering Issues Software Engineering Concepts System Specifications Procedural Design Object-Oriented Design System Testing.
Cross-Layer Scheduling for Power Efficiency in Wireless Sensor Networks Mihail L. Sichitiu Department of Electrical and Computer Engineering North Carolina.
What’s Ahead for Embedded Software? (Wed) Gilsoo Kim
Chapter 1 Basic Concepts of Operating Systems Introduction Software A program is a sequence of instructions that enables the computer to carry.
General requirements for BES III offline & EF selection software Weidong Li.
Lesson 1 1 LESSON 1 l Background information l Introduction to Java Introduction and a Taste of Java.
System Architecture Directions for Networked Sensors.
Energy-efficient Scheduling policy for collaborative execution in mobile cloud computing INFOCOM '13.
Efficient Opportunistic Sensing using Mobile Collaborative Platform MOSDEN.
CSCI/CMPE 4334 Operating Systems Review: Exam 1 1.
Evaluation of Advanced Power Management for ClassCloud based on DRBL Rider Grid Technology Division National Center for High-Performance Computing Research.
Introduction to Operating Systems Concepts
Marilyn Wolf1 With contributions from:
SDN challenges Deployment challenges
Mobile Operating System
Software Design Mr. Manoj Kumar Kar.
Design and Implementation of Spacecraft Avionics Software Architecture based on Spacecraft Onboard Interface Services and Packet Utilization Standard Beijing.
Introduction to cosynthesis Rabi Mahapatra CSCE617
Presented By: Darlene Banta
Implementation support
Mark McKelvin EE249 Embedded System Design December 03, 2002
An XML-based System Architecture for IXA/IA Intercommunication
Implementation support
Presentation transcript:

Power Modeling and Power Management Framework Dexin Li December 2002

Outline Background Power management architecture Component-level power modeling System-level power simulation Energy optimization Preliminary results

Background Challenges on power management in complex embedded systems –Interaction and cooperation among multiple components –Existence of inter-component dependencies –Complicated local mode transitions for a global mode change –High cost for a global mode change Scope of the work –Purpose a software architecture dealing with power management complexity –Model components and component modes with dependency –Generate feasible sequence of mode transitions –Optimize the energy for the cost of mode changes

Power Management Architecture Separation of policy-maker and component details –Modular, retargetable, –Lessen the burden of policy-maker Hierarchical component model –Capable of handling complex systems Modeling of inter-component dependency –Enabling complicated mode transition optimization Online or offline power simulation –Helpful in obtaining feasible mode transitions and optimizing system energy

Execution Model of Power Management Hardware device Operating System Middle-ware Application Power manager We assume this model in the rest of the presentation.

Top-Level Power Manager Power Manager Requirements Device status OS/MW status System spec Component library Power simulator External policy App. status Interdependency Task constraint Power/battery status Control command app os hw Message to app. Message to os

Decoupled Power Management Architecture Diagram

Policy Handler Responsibilities –monitor system status –Make system-level power management policies –Send mode change commands System interaction –Accept service requests from applications –Accept system requirement or environment input from middle-ware –Obtain system status from middle-ware or OS –Send mode change commands to Component Manager

Component Manager Command Interpreter –Receive system command from policy handler –Translate system command into local command Core algorithms –Generate sequence of mode transitions –Optimize it for energy reduction –Send results to device drivers –Feed results to simulation engine

Component Manager Diagram MTG 1 MTG 2 MTG n Mode Dependency Model Device status System status Command Interpreter Core Algorithms Policy Handler Device driver Component Manager

Component Modeling Example: Power Amplifier Device status System status Command Interpreter All-pair-shortest-path Algo. (Floyd-Warshall) Component manager for PA offsb tx rx f1: fsm_lb offsb tx rx f2: fsm_hb offon f3: temp_lb offon f4: temp_hb F1.sb->f2.off F1.tx->f2.off F1.rx->f2.off F2.sb->f1.off F2.tx->f1.off F2.rx->f1.off F1.sb->f3.on F1.tx->f3.on F1.rx->f3.on F2.sb->f4.on F2.tx->f4.on F2.rx->f4.on Dependency model

Component Manager Example: Power Amplifier offsb tx rx f1: fsm_lb offsb tx rx f2: fsm_hb Current state: low band transmit System command: change to high band receive State transition sequence: f1.tx -> f1.sb -> f1.off f2.off -> f2.sb -> f2.rx f4.off -> f4.on offon f4: temp_hb Optimized for energy reduction: f1.tx -> f1.sb -> f1.off f4.off -> f4.on f2.off -> f2.sb -> f2.rx Energy: 3.88J Energy: 1.88J

System-level Mode Simulation Responsibility –Receive sequence of mode transitions –Simulate mode transitions –Parallelize possible mode transitions –Obtain time and energy information –Send the information back to Policy Handler Requirement –Detailed timing information for synchronization –Hierarchical component management –Parallelism among different components –Interface to other power management units

Simulation Engine Diagram

Energy Optimization Objective –Optimize energy consumption for mode changes Techniques –Shortest-path algorithm to optimize for one mode change on one component –Topological sorting to power up high power component as late as possible –Mode simulation to parallelize certain mode changes

Modeling a JTRS Channel Complex system composition –13 power manageable components –Multiple power modes on each component Complicated inter-component dependencies –derived from system specification and application requirements Complicated mode changes –Involve more than one component for a mode change –Take a period of time comparable to the service time

Experimental Results Test cases: –Case 1: all power-up –Case 2: two modes: on and off –Case 3: 5 system modes without our model –Case 4: 5 system modes with our modes Results: –Saves 50%-80% energy when service time is comparable to overhead time –Saves 20%+ energy when service time is 1000x larger than overhead time

Conclusion A new software architecture supporting efficient power management in complex systems Component power model enabling power management in complex systems with dependency Multiple optimization techniques reducing energy cost for global mode changes System-level simulation providing optimized component-level details to global policy-maker

End Slide

Execution Model of the Power Manager Not clear what execution model is best/feasible/practical … Hardware device Operating System Middle-ware Application Power manager

Device Broker MTG 1 MTG 2 MTG n Mode Dependency Model Device status System status Command receiver/ interpreter Broker core System power manager Device driver

Mode Transition Graph offstb tx rx

Execution Model of the Power Manager Not clear what execution model is best/feasible/practical... Hardware device Operating System Middle-ware Application Power manager

Execution Model of the Power Manager Not clear what execution model is best/feasible/practical … Hardware device Operating System Middle-ware Application Power manager

Execution Model of the Power Manager Not clear what execution model is best/feasible/practical … Hardware device Operating System Middle-ware Application Power manager

More Detailed Power Manager Composition Power Manager (core) Component library Simulation Engine Sys. specification App. Requirements App/task Constraints Power management policy Device driver Device model Device broker I Device status Device driver Device model Device broker II Device status Sys/app status