1 The Problem of Power Consumption in Servers L. Minas and B. Ellison Intel-Lab In Dr. Dobb’s Journal, May 2009 Prepared and presented by Yan Cai Fall.

Slides:



Advertisements
Similar presentations
Sabyasachi Ghosh Mark Redekopp Murali Annavaram Ming-Hsieh Department of EE USC KnightShift: Enhancing Energy Efficiency by.
Advertisements

Performance, Energy and Thermal Considerations of SMT and CMP architectures Yingmin Li, David Brooks, Zhigang Hu, Kevin Skadron Dept. of Computer Science,
MULTICORE PROCESSOR TECHNOLOGY.  Introduction  history  Why multi-core ?  What do you mean by multicore?  Multi core architecture  Comparison of.
IT Equipment Efficiency Peter Rumsey, Rumsey Engineers.
Thin Servers with Smart Pipes: Designing SoC Accelerators for Memcached Bohua Kou Jing gao.
Application Models for utility computing Ulrich (Uli) Homann Chief Architect Microsoft Enterprise Services.
Keeping Hot Chips Cool Thermal Management for Green Computing Yang Ge Professor Qinru Qiu.
Room: E-3-31 Phone: Dr Masri Ayob TK 2123 COMPUTER ORGANISATION & ARCHITECTURE Lecture 4: Computer Performance.
Datacenter Power State-of-the-Art Randy H. Katz University of California, Berkeley LoCal 0 th Retreat “Energy permits things to exist; information, to.
Energy Efficient Prefetching with Buffer Disks for Cluster File Systems 6/19/ Adam Manzanares and Xiao Qin Department of Computer Science and Software.
Exploring The Green Blade Ken Lutz University of California, Berkeley LoCal Retreat, June 8, 2009.
Princess Sumaya Univ. Computer Engineering Dept. د. بســام كحـالــه Dr. Bassam Kahhaleh.
Energy Model for Multiprocess Applications Texas Tech University.
Virtual Network Servers. What is a Server? 1. A software application that provides a specific one or more services to other computers  Example: Apache.
Intel  modular server building blocks ( built on Intel  Multi-Flex Technology ) Intel  modular server building blocks ( built on Intel  Multi-Flex.
F1031 COMPUTER HARDWARE CLASSES OF COMPUTER. Classes of computer Mainframe Minicomputer Microcomputer Portable is a high-performance computer used for.
University of Karlsruhe, System Architecture Group Balancing Power Consumption in Multiprocessor Systems Andreas Merkel Frank Bellosa System Architecture.
Data Centre Power Trends UKNOF 4 – 19 th May 2006 Marcus Hopwood Internet Facilitators Ltd.
Folklore Confirmed: Compiling for Speed = Compiling for Energy Tomofumi Yuki INRIA, Rennes Sanjay Rajopadhye Colorado State University 1.
Thermodynamic Feasibility 1 Anna Haywood, Jon Sherbeck, Patrick Phelan, Georgios Varsamopoulos, Sandeep K. S. Gupta.
OPTIMAL SERVER PROVISIONING AND FREQUENCY ADJUSTMENT IN SERVER CLUSTERS Presented by: Xinying Zheng 09/13/ XINYING ZHENG, YU CAI MICHIGAN TECHNOLOGICAL.
Multi Core Processor Submitted by: Lizolen Pradhan
Lecture 03: Fundamentals of Computer Design - Trends and Performance Kai Bu
Cloud Computing Energy efficient cloud computing Keke Chen.
Last Time Performance Analysis It’s all relative
BY:- Ch.Nabeel Ahmed Superior University Grw Campus
The 4 functions of a computer are 1.Input 2.Output 3.Storage 4.Processing.
CERN - IT Department CH-1211 Genève 23 Switzerland t Tier0 database extensions and multi-core/64 bit studies Maria Girone, CERN IT-PSS LCG.
Multi-core Programming Introduction Topics. Topics General Ideas Moore’s Law Amdahl's Law Processes and Threads Concurrency vs. Parallelism.
C OMPUTER O RGANIZATION AND D ESIGN The Hardware/Software Interface 5 th Edition Chapter 1 Computer Abstractions and Technology Sections 1.5 – 1.11.
Future Server and Storage Technology Brian Minick, Infrastructure Design Leader - GE.
Challenges towards Elastic Power Management in Internet Data Center.
Virginia Information Technologies Agency Green Technology Initiatives
Thermal-aware Issues in Computers IMPACT Lab. Part A Overview of Thermal-related Technologies.
Data Replication and Power Consumption in Data Grids Susan V. Vrbsky, Ming Lei, Karl Smith and Jeff Byrd Department of Computer Science The University.
Software Architecture for Dynamic Thermal Management in Datacenters Tridib Mukherjee Graduate Research Assistant IMPACT Lab ( Department.
Critical Power Slope: Understanding the Runtime Effects of Frequency Scaling Akihiko Miyoshi †,Charles Lefurgy ‡, Eric Van Hensbergen ‡, Ram Rajamony ‡,
PERFORMANCE STUDY OF BIG DATA ON SMALL NODES. Ομάδα: Παναγιώτης Μιχαηλίδης Αντρέας Σόλου Instructor: Demetris Zeinalipour.
MULTICORE PROCESSOR TECHNOLOGY.  Introduction  history  Why multi-core ?  What do you mean by multicore?  Multi core architecture  Comparison of.
Performance and Energy Efficiency Evaluation of Big Data Systems Presented by Yingjie Shi Institute of Computing Technology, CAS
Energy Efficient Data Centers Update on LBNL data center energy efficiency projects June 23, 2005 Bill Tschudi Lawrence Berkeley National Laboratory
1 Thermal Management of Datacenter Qinghui Tang. 2 Preliminaries What is data center What is thermal management Why does Intel Care Why Computer Science.
Datacenter Energy Efficiency Research: An Update Lawrence Berkeley National Laboratory Bill Tschudi July 29, 2004.
GreenCloud: A Packet-level Simulator of Energy-aware Cloud Computing Data Centers Dzmitry Kliazovich ERCIM Fellow University of Luxembourg Apr 16, 2010.
Accounting for Load Variation in Energy-Efficient Data Centers
E-MOS: Efficient Energy Management Policies in Operating Systems
BY ZENIFA SHARMIN ALI (970282) ARNAB MALLICK (972127) ASHUTOSH MALI (976524) ASHUTOSH RAJ (988061) DEBANJAN KUNDU (986400) COMPUTER, LAPTOP SERVER AND.
CS203 – Advanced Computer Architecture
Copyright © 2010 Hitachi Data Systems. All rights reserved. Confidential – NDA Strictly Required Hitachi Storage Solutions Hitachi HDD Directions HDD Actual.
By Harshal Ghule Guided by Mrs. Anita Mahajan G.H.Raisoni Institute Of Engineering And Technology.
Computer Hardware Components. Cases There are desktop cases and tower cases.
History of Computers and Performance David Monismith Jan. 14, 2015 Based on notes from Dr. Bill Siever and from the Patterson and Hennessy Text.
Computer 3 JEOPARDY Bobbie, Sandy, Trudy.
CS203 – Advanced Computer Architecture
Computer Hardware.
Green cloud computing 2 Cs 595 Lecture 15.
CS111 Computer Programming
Morgan Kaufmann Publishers
Architecture & Organization 1
Green Software Engineering Prof
Unit 2 Computer Systems HND in Computing and Systems Development
Server Innovation Accelerates IT Transformation
IT Equipment Efficiency
Hui Chen, Shinan Wang and Weisong Shi Wayne State University
Architecture & Organization 1
IT Equipment Efficiency
The Greening of IT November 1, 2007.
SERVER INNOVATION ACCELERATES IT TRANSFORMATION
The University of Adelaide, School of Computer Science
Installing A Graphics Card
Presentation transcript:

1 The Problem of Power Consumption in Servers L. Minas and B. Ellison Intel-Lab In Dr. Dobb’s Journal, May 2009 Prepared and presented by Yan Cai Fall 2009 Green Computing

Motivation Server energy consumption has increased from 50w to 250w since 2000 Energy cost will exceed the server cost if this trend does not change 2 Source: IDC, Scaramella 2006

Outline Motivation Where heat comes from How cooling is achieved How power is consumed Conclusions 3

Where heat comes from Server form factor Server power consumption is affected by server form factor, including  the individual configuration,  the heat and thermal environment  the workload being processed  … 4 Server form factor Pedestal servers 2U rack servers 1U rack servers Blade servers 1U Pedestal 2U Blade

Where heat comes from The amount of heat, (Q), generated by an integrated circuit is a function of  The efficiency of the components' design (e)  The technology used in its manufacturing process (t)  The frequency and voltage at which it operates (f,v) Over-clocking also generates a great amount of heat 5

Outline Motivation Where heat comes from How cooling is achieved How power is consumed Conclusions 6

How cooling is achieved Two common methods  Heat sinks  Fans 7 Natural convection heat sink (Source: Wikipedia, 2008)

Outline Motivation Where heat comes from How cooling is achieved How power is consumed Conclusions 8

How power is consumed ServersCost# of Processors Volume$25,0001 ~ 2 Mid-range$25,000 ~ $499,9992 ~ 4 High-end≥ $500,000≥ 8 9 Estimated Average Power Use (W) per Server, by Server Class, 2000 to 2006 (Source: Koomey J 2007b Estimating Total Power Consumption by Servers in the US and the World. Oakland, CA: Analytics Press) Estimated Average Power Use (W) per Server Google container Data Center

Total power consumption Power consumption of a Quadcore Intel Xeon server 10 Server Power Consumption (Source: Intel Labs, 2008)

Total power consumption (cont’d) Power consumption of components 11 ComponentsPower consumption Multi-core CPU45 ~ 200 w DIMM5 ~ 21 w Power supply efficiency lossSimilar with memory p.s. 2 PCI Slots50 w 2 ~ 4 Hard drives24 ~ 48 w

Total power consumption (cont’d) Total power consumption Estimation of power consumption 12 CPU Utilization and Power Consumption (Source: Blackburn 2008)

Memory power consumption The memory demand is growing in the future, because  More processors on chip,  Increasing use of virtualization,  More memory intensive search applications (Google and Facebook) The memory power consumption is growing in the future  The more the memories, the larger the power consumption  The faster the memories, the larger the power consumption 13

Memory power consumption (cont’d) Memory Power Comparison 14 RDIMM Memory Power Comparison (Source: Intel Platform Memory Operation, 2007)

Memory power consumption (cont’d) Power consumption by vendors and configurations 15 RDIMM Power Consumption by Vendor and Configuration (Sources: Publicly-available datasheets from each vendor, 2008)

Memory power consumption (cont’d) 64GB system power consumption 16 Future DIMM Power Consumption by Frequency, Configuration, and Capacity (Source: Intel Platform Memory Operation, 2008)

Memory power consumption (cont’d) How to reduce memory energy consumption  By doing thermal analysis, embedding thermal sensors and throttling 17 1U Server Architecture (Source: Intel Labs, 2006)

Memory power consumption (cont’d) Memory throttling mechanisms  Closed loop thermal throttling (CLTT) Closed loop, rely on thermal sensors  Open loop throughput throttling (OLTT) Open loop, work based on bandwidth count Memory throttling degrades performance 18

Power supply efficiency loss Power supply efficiency vs. utilization level 19 Server Power Supply Efficiency Test Report (Source: Lawrence Berkeley National Laboratory, 2004)

Storage power consumption Average hard disk power consumption for average operations Average hard disk power consumption for IO intensive operations 20

Conclusions High-end servers consume more power, but might be less efficient in terms of power consumption At present, CPU still consumes the most power, compared to other components In the future, memory can consume more power than CPU Considering a significant amount of power loss, the efficiency of power supplies is also very important The power consumption by hard drives is closely related to the workload the server is processing 21