Laptop Computers How Much Energy Do They Use,

Document Sample
Laptop Computers How Much Energy Do They Use, Powered By Docstoc
					                                                                                                             September 2003




      Laptop Computers:
      How Much Energy Do They Use,
      and How Much Can We Save?
      Highlights of project managed by Noah Horowitz, Sr. Scientist
      Prepared by Suzanne Foster and Chris Calwell of Ecos Consulting

Executive Summary                            The Savings Opportunity                    Monitor choice greatly impacts the overall
Today in the United States there are over    While the fastest desktop computers        system energy usage. For example, simply
200 million computers in use in business     continue to outpace the fastest laptops    switching from a CRT to the increasingly
and residential settings. The annual         in absolute speed and processing power,    common LCD (liquid crystal display), yields
national energy use of computers and         it is now very common to find affordable   the desktop computer user savings of
their associated monitors is approximately   desktop and laptop computers with          more than 100 kWh/yr. Likewise, much
85 billion kWh per year, or approximately    roughly comparable performance,            of the energy savings potential from a
2.8% of total US electricity use. Due to     especially when running the common         laptop is negated if it is connected to an
their portability, improved performance,     software applications found in most        external CRT monitor. This figure also
and reduced purchase costs, laptop           offices. As illustrated below, today’s     shows that even with power manage-
computers are becoming an increasingly       average desktop computer with a            ment enabled, the majority of the
popular choice for residential and com-      cathode ray tube (CRT) monitor uses        energy consumption happens while the
mercial consumers alike. Laptops employ      more than five times as much energy as     computer is in the On mode, not in the
the latest developments in computer          a laptop (570 kWh/yr vs. 100 kWh/yr).      various low power modes like Monitor
technology due to the need to prevent        Therefore, replacing a desktop system      Sleep, Hardware Sleep, and Off.
overheating and to maximize battery life.    with a laptop system could save about
                                             470 kWh/yr - nearly $40 annually.
To date, there has been little work done
to assess and compare the energy
efficiency of laptop computers. To this
end, NRDC and its consultant, Ecos
have prepared this summary and a larger
full report. The purpose is to assess the
approximate energy efficiency differences
between laptops and desktops and
among various models of each type in
order to determine how much energy
the U.S. might save by switching to more
efficient designs. In the process, we
discuss various ways to characterize
computer efficiency on a component
basis and system basis, and conclude with
recommendations for future policy and
research activities in this field.
Environmental Impacts                           computers and other consumer                  while the rest is likely due to differences
The table below puts some of the                electronics products. The best designs        in the technologies employed. Some
potential energy and environmental              could maintain efficiencies nearly that       displays, for example, provide better off-
impacts into perspective, demonstrating         high over most of their operating range,      angle viewing or more accurate rendering
the significant benefits that can be            while others declined in efficiency fairly    of colors and moving images than others,
attained from particular shifts in the          dramatically at partial load, where the       but may consume more power to do so.
current computer marketplace. The               system operates a substantial portion
section below provides an initial attempt       of the time.                                  CPUs - The three largest central process-
                                                                                              ing unit (CPU) manufacturers for laptop
      Scenario              Energy Saved           Dollars Saved        Tons of CO2 Saved
                                                                                              computers — Intel, AMD, and Transmeta
 25% of US desktops                                                                           — all have recently released chips that
with CRTs are replaced     20 billion kwh/yr       $1.6 billion/yr       13 million tons/yr
     with laptops                                                                             improve computer energy efficiency
 25% of US desktop                                                                            significantly. Most of these designs are
  systems shift from       10 billion kwh/yr        $.8 billion/yr       6 million tons/yr    “mobile” processors only, meaning their
     CRTs to LCDs
                                                                                              efficiency features (the ability to run at
to quantify the potential savings that          Battery Charging Systems - Our                lower clock frequencies and voltages
can be achieved for laptops through             measurements did not reveal any major         during periods of inactivity) are specifically
further improvements in their design            efficiency differences in battery charging    intended for laptop computers. This
and technology.                                 systems across the various laptop models      includes Intel’s Pentium M (Centrino
                                                tested, though differences in battery         system) chips and AMD’s “PowerNow!”-
Key Efficiency Differences                      capacity and run-time were significant.       equipped Athlon processors. Transmeta’s
In order to investigate efficiency              The need for portability appears to           Crusoe is predominantly a laptop proces-
differences among laptop computers,             keep most such battery components fairly      sor, but has also been utilized in NEC’s
we used two different approaches:               similar. Smart charging technology allows     new PowerMate Eco desktop system. In
                                                batteries to automatically inform the         most cases, the manufacturer’s motivation
1) comparing individual components,             charger regarding their temperature and       for selecting the more efficient chip is to
which included the power supply,                state of charge. On average, about 20%        maximize the amount of time a user can
battery charging system, display,               more AC energy is required to operate a       use the laptop between battery charges.
central processing unit, and power              computer in battery mode and then
management software.                            recharge it, compared to simply leaving       Power Management - Power manage-
                                                it plugged in. By contrast, we have found     ment software allows a computer to
2) comparing overall product speed              the battery charging systems used in          reduce power to various components
and performance per unit of energy              other consumer products to be much            after periods of inactivity. The largest
consumed while utilizing system                 less efficient.                               savings typically result from dimming or
performance benchmarking software.                                                            switching off the screen, reducing proces-
                                                Displays - Laptops’ LCD displays vary         sor speed, shutting down the hard drive,
Power Supplies - Our measurements               widely in the amount of information           or putting the entire computer into
showed that laptop power supplies are           (pixels) they can display per watt of         “sleep” mode after longer periods of
remarkably efficient, typically achieving       power consumed. Some of this is               time. This sleep capability varies widely
nearly 90% efficiency at peak load. This        attributable to differences in screen         across various operating systems,
is far higher than the efficiencies typically   brightness, size, and performance,
found in power supplies used in desktop
computer manufacturers, and component                           for each component overlap somewhat               chosen because of their claims to repre-
types, and is frequently not fully enabled                      so cannot be simply added together, but           sent “normal” office user performance,
by the user. In most operating systems,                         they do approximate the energy benefits           while PC Mark was chosen because it
maximum performance is synonymous                               that are associated with improving the            is easy to use. Unfortunately, we were
with AC state and slower performance is                         efficiency of each component individually.        unable to gather meaningful energy
synonymous with battery state, meaning                                                                            consumption values that related to the
that large savings opportunities can go                         Applying Benchmarking Software                    performance score for PC Worldbench4,
untapped when computers are plugged                             to Assess System Performance                      and as a result, only MobileMark and
into AC power.                                                  Looking at the individual components can          PCMark were used to generate system
                                                                make comparisons unnecessarily compli-            efficiency metrics.
                                                                cated and not entirely accurate, since the
                                                                laptop components operate together as a           The resulting laptop system efficiencies,
                                                                system. Also, the simple measures of              as defined using MobileMark software,
                                                                battery life frequently published in              are listed in the table on the following
                                                                computer magazines are not meaningful             page. These data represent initial efforts
                                                                comparisons of the relative efficiencies of       by Ecos to assess the validity of this type
                                                                two laptops. One laptop can achieve a             of system level approach and other
                                                                longer battery life than the other by             benchmarks could change the relative
This power management window, available to the user through     simply running slower or having a larger          performance ranking of the laptops
Windows 2000 software, allows the user to set different power
management functions based on whether the computer is plugged   battery. Using benchmarking software              tested. Nevertheless, these data indicate
in (AC state) or running on batteries (Battery state).
                                                                tools to compare functional performance           that large system efficiency differences
What Is the Impact of More                                      per unit of AC energy consumed can help           exist among laptop computers with
Efficient Laptop Components?                                    to improve the fairness and relevance of          different configurations (larger scores
In the table below, the efficiency differ-                      efficiency comparisons across a range of          indicate greater efficiencies). Even within
ences of various components are summa-                          laptop and desktop computers. This type           one form factor (thin and light) we see
rized. Also given are the estimates of how                      of system metric measures the efficiency          that the highest score (6.3) is more than
much energy would be saved if the most                          of the interaction of all the components.         three times the lowest score (1.9).
efficient components were adopted.
The savings are quantified in two                               Because there is currently no benchmark-          We also employed PCMark to compare
different duty cycle scenarios: the Road                        ing software that is ideal for an energy          desktops to laptops. Efficiency was
Warrior scenario, which refers to a laptop                      efficiency metric, Ecos evaluated three           calculated in a similar way as the
as it is used by a business traveler, and                       different benchmarking tools: BAPCO’s             MobileMark benchmark: total perform-
the Desktop Replacement scenario, which                         MobileMark 2002, Futuremark’s PCMark              ance score (CPU+memory+hard drive)
refers to a laptop that remains plugged in                      2002, and PC World’s PC WorldBench4.              divided by the AC watt-hours
most of the time. The potential savings                         MobileMark and PC WorldBench4 were                consumed during the test.
                                                                                         National Annual Energy Savings:       National Annual Energy Savings:
         Component                              Possible Efficiency Improvement                Road Warrior (GWh)               Desktop Replacement (GWh)

                                                      From 64,000 pixels/watt
            Display
                                                       to 128,000 pixels/watt
                                                                                                      260                                   550
                                            From partial load efficiency of 56% to 85%
       Power Supplies
                                              From full load efficiency of 80% to 90%
                                                                                                      210                                   520
                                                  From Intel P3 Max Performance
              CPU
                                                     to P3 Battery Optimized
                                                                                                      180                                   520
                                                 From 50% of laptops PM enabled
   Power Management
                                                    to 70% laptops PM enabled
                                                                                                       20                                   220
                                                    From 80% to 85% efficiency
       Battery System                                                                                  30                                    9
                                               Laptop                 CPU Power           MobileMark          MobileMark            Measured Energy               System
                                             Tested by               Management           Performance         Battery Life         to Charge Battery             Efficiency2
                                          Ecos Consulting              Enabled?               Score             (hours)             (AC Watt-hours)            (Performance/
                                                                                                                                                                   Watt)

                                              IBM T23
                                           (Intel P3 Mobile)                No                  111                 3.3                     58.8                       6.3
                                            Thin and Light

                                              IBM T40
                                            (Intel Centrino)                Yes                  95                 4.2                     66.1                       6.0
                                             Thin and Light

                                           Sharp MM-10
                                          (Transmeta Crusoe)                Yes                  60                 2.5                     35.2                       4.3
The range of scores for the                  Ultra Portable
laptops was 532 to 3696 and            Fujitsu S-Series Lifebook
for desktops was 295 to 519,               (AMD Athlon 4)                   Yes                  94                 2.4                     58.6                       3.9
                                            Thin and Light
where a higher score indicates
higher efficiency. Note that the               MiTAC
                                           (AMD Athlon 4)                   No                   66                 2.2                     77.1                       1.9
most efficient desktop we                   Thin and Light

measured, the NEC Powermate             Toshiba Tecra 8100
Eco PC, achieved a score                      (Intel P3)                   NA1                   50                 2.4                     67.0                       1.8
                                        Desktop Replacement
comparable to that of the
                                   Using this type of efficiency metric, a laptop like the IBM T40 can achieve a high efficiency score by attaining a relatively good performance
least efficient laptop we          score while using an average amount of AC energy. Alternatively, the Sharp laptop achieved a better than average efficiency score with a
                                   relatively low performance score because it consumed a comparatively small amount of AC energy.
measured.
                                   1
                                       There is only one power mode for this CPU.
                                   2
                                       System Efficiency = (MobileMark Performance Score) * (MobileMark Battery Life) / (Measured Energy to Charge Battery)




    CONCLUSIONS
    Our preliminary research suggests three important conclusions:

G Laptops offer the potential for major              G While there are clearly significant                            G Specifications such as ENERGY STAR®
    energy savings relative to desktops.                   energy efficiency differences in the                             that currently recognize efficient
    The best laptops are at least 5                        components of computers, those                                   computers in the marketplace should
    times more energy efficient than                       technologies change rapidly and                                  be revised to include consideration
    the worst desktop systems                              interact closely, making it difficult to                         of on mode energy use – the most
    (computer and CRT monitor).                            drive overall improvements in system                             important single fraction of overall
    However, highly efficient laptop                       efficiency with component-based                                  energy use. Based on our preliminary
    components could be readily                            specifications. Employing instead                                measurements, separate specifica-
    incorporated into desktop designs,                     benchmarking software and an AC                                  tions for laptops and desktops are
    preserving the basic form factor                       watt-hour meter to capture overall                               warranted. A similar methodology
    and functionality of a desktop while                   product efficiency in On mode offers                             should be employed in each,
    saving energy and space and                            key advantages. It allows manu-                                  however, so buyers understand
    reducing noise from cooling fans.                      facturers to choose combinations of                              the savings they could achieve by
    However, the present trend is the                      individual components and software                               purchasing not just a more efficient
    reverse, with each new desktop                         to achieve the greatest energy savings                           desktop and screen, but a highly
    computer model incorporating faster,                   at the lowest cost, and reduces the                              efficient laptop instead.
    higher power CPUs and video cards.                     need for frequent updates to the
                                                           technical details of an efficiency
                                                           specification. The next step to further
                                                           understand and quantify these
                                                           system level differences is to create,
                                                           with input from stakeholders in the
                                                           computer industry, a benchmark
                                                           specifically for energy efficiency.


                                                                           The research performed to prepare this report was funded by a grant from the U.S. Environmental
                                                                           Protection Agency (EPA) to the Natural Resources Defense Council (NRDC). The views and findings
                                                                     expressed herein are solely those of the authors and do not necessarily state or reflect those of the EPA.
                                                               For more information, contact the Project Manager, Noah Horowitz, at nhorowitz@nrdc.org or (415)777-0220.