Viewing page 1 of 2 pages.
1 2 NextNovember 21, 2006 by Alec Ross
|
Editor Note: Alec Ross, currently enrolled in the CS program at the University of Toronto, usually hangs out in the SPCR newsroom. Working on this story, Alec found so many changes from the original proposal for the new ES computer specification that it quickly ballooned into a full article. Energy conscious readers will find his report very interesting, as will those who like to follow industry politics.
Mike Chin
|
The U.S. Environmental Protection Agency was created in
1970, with the grand mission of "protecting human health and the environment." In 1992, it
introduced the EnergyStar program for computers and monitors, which expanded
over the 1990s to include over 50 additional categories. EnergyStar is a
voluntary program: In exchange for meeting the EPA's efficiency targets,
companies get to use the familiar star logo, and receive favorable treatment
when bidding on government orders.
The previous standards, which came into
effect on July 1, 2000, required only that the computer have an automatic "sleep
mode" during which it could consume up to 30% of it's maximum AC power,
depending on the size of the power supply. As a result, over 98% of computers
shipping today bear the EnergyStar logo. One of the goals of the new
specifications was to reduce that number — in effect, to make certification
something that only the cream of the crop can boast of having. The new Energy Star Computer Specification, Version 4.0, will become effective July 20, 2007.
The new standards have undergone a number of revisions, as reported by Mike Chin
in his article "The State of the Industry,
March 2006: Through Silent Eyes", based on feedback from industry
representatives such as Hewlett Packard, ATI and Intel (all of which are
available
on the EPA's website.) These revisions have been fairly extensive, which is why you
get a whole article instead of a news post. Their net effect? As you may have
guessed, after industry had their say, the certification was loosened and expanded quite a
bit.
First, the big question: How efficient does a computer need to be to qualify for
EnergyStar? Well, it depends what kind of computer we are talking about.
There are no fewer than six
classes of computer under the new EnergyStar. This is the largest single change
from the original plan, which called for one level of efficiency for all
desktops, one for all notebooks, and so on. In the final spec, each different class has different
feature and efficiency requirements. Some, such as dedicated servers, blade
units and palmtops, aren't included at all.
There
are a couple of constants across every type and tier:
- Sleep Mode: Computers are still
required to enter sleep mode automatically after 30 minutes by default (except
for servers, which are generally working 24/7).
- Power Supply Efficiency: External power supplies must
exceed 84% efficiency. More importantly, internal power supplies must have a
power factor over 0.9 at 100% load, and an efficiency rating of at least 80% at
the 20%, 50%, and 100% load levels. This is a major victory for efficiency
and, by association, low noise.
At present, the minimum efficiency for an ATX 2.2
power supply is 68%, and power factor regulation is entirely voluntary in the
US. While Ecos Consulting's 80-Plus program was well-received at SPCR, its
visibility to the mainstream media and consumers was initially low, though
this has changed in the last few months with EnergyStar's adoption of
the standard. With this specification, power supply efficiency has been vaulted
into the spotlight, and the fine folks at Ecos should give themselves a
well-deserved pat on the back. The power factor correction is icing on the cake;
not only will computer power supplies be more efficient, they will also put less of a
strain on electricity grids.
The table below summarizes the EnergyStar Final Computer Specification, Version 4.0, which can be downloaded at http://www.energystar.gov/index.cfm?c=revisions.computer_spec.
|
Summary of EnergyStar 4.0 Specifications
- effective July 20, 2007
|
|
Computer Type
|
Idle Power
|
Sleep Power
|
Standby Power
|
Efficiency
|
Power Factor
|
Notes
|
|
Category A Desktop Derivatives
|
50W
|
4.0W
|
2.0W
|
80%
|
0.9
|
Standard Desktops
|
|
Category B Desktop Derivatives
|
65W
|
4.0W
|
2.0W
|
80%
|
0.9
|
Dual-Core Systems with 1GB RAM
|
|
Category C Desktop Derivatives
|
95W
|
4.0W
|
2.0W
|
80%
|
0.9
|
Gaming/Heavy Multimedia Systems
|
|
Category A Notebooks + Tablets
|
14W
|
1.7W
|
1.0W
|
80%
|
0.9
|
Most Laptops
|
|
Category B Notebooks +
Tablets
|
22W
|
1.7W
|
1.0W
|
80%
|
0.9
|
Laptops w/ 128MB Graphics Memory
|
|
Workstations
|
Special
|
Special
|
Special
|
80%
|
0.9
|
Commercial Workstations
|
There are a few points in the table above that bear additional clarification:
"Desktop
derivatives" include actual desktops, desktop-derived servers, integrated
computers and game consoles.
Workstations have their own special set of
rules based on "Typical Electricity Consumption", or TEC, which is 35% of the
maximum power supply output plus 1.75 watts per hard drive. The maximum idle,
sleep and standby power ratings are then determined by the TEC; if the formula
(0.7*Idle +0.2*Sleep +0.1*Standby) is less than the TEC, the model qualifies.
This gives integrators some wiggle room- a slightly worse efficiency at idle can
be compensated by better efficiency while sleeping. Compared to desktop
derivatives, this seems like a pretty low target — a computer with a 400W power
supply and two hard drives has a TEC rating of 143.5W. When you consider all the
equipment on the list to qualify as a workstation, however — ECC memory, discrete
graphics, multiple processor sockets — the reason for the high power allowance
becomes clear.
How tough are these targets?
The category A desktop target is fairly demanding by current standards. The only modern PCs SPCR has evaluated that idle below
50W are VIA Mini-ITX systems, or desktop systems using mobile CPUs.
SPCR's Athlon64-based lab computer draws 51W when idling, just slightly
off-pace. According to most review sites, Intel's flagship Core 2 Duo
draws more power at idle, which likely puts it out of the running. There
are, of course, a number of other options: AMD's Athlon64 EE HTPC variant will
afford a bit of headroom by using only 35W at peak, Intel has a slew of mobile
chips all the way down to 9W, and VIA's systems would nearly all fit in under
the limit — not to mention Transmeta's Efficeon, AMD's Geode, and so on. Many of
these would make for a very efficient system, at least without too much in the
way of RAM, graphics power, and so on.
Power supply sizing will also play a factor in meeting the spec. Most power
supplies drop off drastically in efficiency below 20% load, and if the system is
to idle at 50W, a 400W or 500W power supply just isn't going to be efficient
enough. In the same vein, and as MikeC mentioned in his first article,
motherboard efficiency is suddenly much more important. If it's optimized to
deliver 130W to the CPU, the motherboard will likely have trouble with
efficiency when idling at 10W or 20W.
| Help support this site, buy from one of our affiliate retailers! |
|
