Measured wattage of atom system

[D G Teed-2]

As I've mentioned here last week, I'm installing
Debian on an Atom based system to replace
a PIII and Sun Ultra 1 at home.

Here are some wattage comparisons measured with
Kill A Watt.

PIII dual 550 Mhz with 2 320 GB IDE seagate drives
Idle: 53 watts
Typical load: 86 watts

Sun Ultra 1 167 Mhz with 2 18 GB 10K SCSI drives
Idle: 94 watts
Typical load: 104 watts

Atom 1.6 Ghz with 2 320 GB SATA seagate drives
Idle: 42 watts
Typical load: 47 watts

The Atom motherboard and processor is a funny setup.
The heatsink and fan are over the main chipset, and there
is a tiny passive heatsink over the CPU.  Quiet the twist
on what we're used to seeing.

Part of the reason I kept the PIII system for so long was
that it just worked, and it was cheaper on power than
any upgrade path previously available.

--Donald
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Jason Kenney]

So assuming you keep your systems on 24/7,

at the current (ie more now than in the past) cost of electricity,
10.67 cents/kW*hour

every 100W uses 25.6 cents of electricity a day.

You're saying that probably at the very most a quarter a day of
electricity was a significant factor in choosing what hardware you
used?

Really?!


Jason


On Tue, Sep 16, 2008 at 12:32 AM, D G Teed <donald.teed@...> wrote:

> As I've mentioned here last week, I'm installing
> Debian on an Atom based system to replace
> a PIII and Sun Ultra 1 at home.
>
> Here are some wattage comparisons measured with
> Kill A Watt.
>
> PIII dual 550 Mhz with 2 320 GB IDE seagate drives
> Idle: 53 watts
> Typical load: 86 watts
>
> Sun Ultra 1 167 Mhz with 2 18 GB 10K SCSI drives
> Idle: 94 watts
> Typical load: 104 watts
>
> Atom 1.6 Ghz with 2 320 GB SATA seagate drives
> Idle: 42 watts
> Typical load: 47 watts
>
> The Atom motherboard and processor is a funny setup.
> The heatsink and fan are over the main chipset, and there
> is a tiny passive heatsink over the CPU.  Quiet the twist
> on what we're used to seeing.
>
> Part of the reason I kept the PIII system for so long was
> that it just worked, and it was cheaper on power than
> any upgrade path previously available.
>
> --Donald
> _______________________________________________
> nSLUG mailing list
> nSLUG@...
> http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug
>

_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Stephen Gregory-2]

Jason Kenney wrote:
>
> You're saying that probably at the very most a quarter a day of
> electricity was a significant factor in choosing what hardware you
> used?

That is $90 a year. The Atom system will pay for itself in under two
years. And it is likely faster.


> Really?!

yes. Why through away money?

-- sg

_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[William Lachance]

On Tue, 2008-09-16 at 06:34 -0400, Stephen Gregory wrote:
>
> > You're saying that probably at the very most a quarter a day of
> > electricity was a significant factor in choosing what hardware you
> > used?
>
> That is $90 a year. The Atom system will pay for itself in under two
> years. And it is likely faster.

The residential rate was just set to be increased by 10% for 2009, so
make that $99 a year. :) Given both the soaring price of coal and the
push towards even more expensive renewables for power generation, I
think one can expect the savings from a lower wattage computer to be
even greater in 2010 and beyond.

All this being said, the energy consumption of a computer is generally
dwarfed by that of appliances such as fridges and driers. If saving
power is your primary goal, look at replacing those first. :)
--
William Lachance <wrlach@...>


_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Jason Kenney]

Because it's not "throwing away money" if you are getting value for
it.   It just strikes me as very odd that someone who is into
computers so much would place such a low value on their own
experience.


On Tue, Sep 16, 2008 at 7:34 AM, Stephen Gregory <nslug@...> wrote:

> Jason Kenney wrote:
>>
>> You're saying that probably at the very most a quarter a day of
>> electricity was a significant factor in choosing what hardware you
>> used?
>
> That is $90 a year. The Atom system will pay for itself in under two
> years. And it is likely faster.
>
>
>> Really?!
>
> yes. Why through away money?
>
> -- sg
>
> _______________________________________________
> nSLUG mailing list
> nSLUG@...
> http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug
>

_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Stephen Gregory-2]

On Tue, Sep 16, 2008 at 12:03:55PM -0300, Jason Kenney wrote:
> Because it's not "throwing away money" if you are getting value for
> it.

Agreed.

> It just strikes me as very odd that someone who is into
> computers so much would place such a low value on their own
> experience.

If the Atom is being used to replace two old systems it is likely
overkill for what is required. When all the requirements for a system
are met by an Atom system then the only criteria left to judge the
systems is power consumption.1

It strikes me as very odd the someone who is into computers so much
would not want a small lower power home server.

--
sg
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Stephen Gregory-2]

On Tue, Sep 16, 2008 at 12:32:16AM -0300, D G Teed wrote:
>
> Atom 1.6 Ghz with 2 320 GB SATA seagate drives
> Idle: 42 watts
> Typical load: 47 watts

Did you get a special low wattage supply for this system? Or did you
just use a generic power supply?

--
sg
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Jason Kenney]

I guess you misunderstood - my comments were directed at the choice of
keeping the PIII until now, not making a switch now.

Frankly I don't think power consumption should ever be a consideration
for a home computer system.  The effort you spend considering it could
be more effectively spent finding other ways to reduce expenses in
your life.  The 100W figure is even extreme, if the difference is only
30W, you're now down to what? 7 cents a day?  I'm pretty you "throw
away" significantly more money than that every time you drive your car
by being a far from optimal driver for example.


Jason

On Tue, Sep 16, 2008 at 12:20 PM, Stephen Gregory <nslug@...> wrote:

> On Tue, Sep 16, 2008 at 12:03:55PM -0300, Jason Kenney wrote:
>> Because it's not "throwing away money" if you are getting value for
>> it.
>
> Agreed.
>
>> It just strikes me as very odd that someone who is into
>> computers so much would place such a low value on their own
>> experience.
>
> If the Atom is being used to replace two old systems it is likely
> overkill for what is required. When all the requirements for a system
> are met by an Atom system then the only criteria left to judge the
> systems is power consumption.1
>
> It strikes me as very odd the someone who is into computers so much
> would not want a small lower power home server.
>
> --
> sg
> _______________________________________________
> nSLUG mailing list
> nSLUG@...
> http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug
>

_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[D G Teed-2]

On Tue, Sep 16, 2008 at 1:51 AM, Jason Kenney <jdkenney@...> wrote:

> So assuming you keep your systems on 24/7,
>
> at the current (ie more now than in the past) cost of electricity,
> 10.67 cents/kW*hour
>
> every 100W uses 25.6 cents of electricity a day.
>
> You're saying that probably at the very most a quarter a day of
> electricity was a significant factor in choosing what hardware you
> used?

The calculation I have for under load is 143 watts difference,
and according to the NS Power calculator that works out to $133/year

Keep in mind, part of the goal in this is finding an upgrade.  The PIII system
maxes out at 1 GB of RAM, which amavisd can take over half, and growing
in each new version.  All other upgrade paths I saw used more electricity than
a PIII generation, so I balked at upgrading for some time.

This is part of the energy conservation process.   I'm also looking to stop
my use of a Cisco switch which eats 40 watts, compared to chaining
small consumer switches which use 1 to 4 watts.  Desktops are only turned on
while in use, rather than left on.  The changes all add up to something
significant compared to where I was, and where I could have been going.

--Donald
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[D G Teed-2]

On Tue, Sep 16, 2008 at 12:23 PM, Stephen Gregory <nslug@...> wrote:
> On Tue, Sep 16, 2008 at 12:32:16AM -0300, D G Teed wrote:
>>
>> Atom 1.6 Ghz with 2 320 GB SATA seagate drives
>> Idle: 42 watts
>> Typical load: 47 watts
>
> Did you get a special low wattage supply for this system? Or did you
> just use a generic power supply?

I bought a new Seasonic 330W, 80 plus efficiency.
I would have liked lower wattage, but couldn't find anything
which was also 80 plus certified.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[D G Teed-2]

I'm going to predict that the Atom is the first generation of a change
we are going to see.  The next version is 64 bit, duo core, higher RAM ceiling,
and some other improvements.  It would be useful for many office desktops
and will be on the market in weeks.

The third generation will likely be for server room applications -
possibly in blades.
In that scenario, you save power twice: once in the server's draw,
and the second time in reduced A/C draw.

Electricity never goes down in price.  The new hardware is cheap (so far).
There is a hosting company in the U.K. which already offers cheaper
hosting on Atom based systems because of this double up on power saving.

AMD may be able to jump in there too, if it acts quickly enough.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[George N. White III]

On Tue, Sep 16, 2008 at 1:51 AM, Jason Kenney <jdkenney@...> wrote:

> So assuming you keep your systems on 24/7,
>
> at the current (ie more now than in the past) cost of electricity,
> 10.67 cents/kW*hour
>
> every 100W uses 25.6 cents of electricity a day.

Virtually all the energy used by a PC ends up as heat.
In NS, for >6 months of the year, the heat is not "waste".
It contributes to the overall heating for your home/office, and
should be subtracted from the cost of other heat sources.

In other words, compared to Silicon Valley, where for 6
months of the year you pay to remove the heat, the economics
of running a PC are much different.

There are other costs/benefits of heat -- most components age more rapidly,
so if the lower energy consumption allows you to keep the components
cooler, you benefit (at least statistically) from improved reliability.
Depending on the surroundings, noise from cooling fans can be a
plus or a minus.

> You're saying that probably at the very most a quarter a day of
> electricity was a significant factor in choosing what hardware you
> used?

--
George N. White III <aa056@...>
Head of St. Margarets Bay, Nova Scotia
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Ian Campbell-9]

On Tue, Sep 16, 2008 at 12:57:48PM -0300, D G Teed wrote:
> I'm going to predict that the Atom is the first generation of a change
> we are going to see.

You're a little late to the party in terms of the drive for lower
power chips. The ULV Pentium M chips drew something like 5W a couple
years ago. If you want to exclude laptop chips, VIA has been sitting
in that niche for years. In the datacenter world look at the HE
opteron chips.


_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Daniel MacKay-2]

> In NS, for >6 months of the year, the heat is not "waste".
> It contributes to the overall heating for your home/office, and
> should be subtracted from the cost of other heat sources.

At Solar Nova Scotia I make this argument on a regular basis, George.  
I don't expect anyone else to get it :-)

When people are arguing about the efficiency of the blowers moving  
air around their house, I ask them where the wasted energy from an  
inefficient blower goes.

The consensus seems to be, "away."

Laws of thermodynamics?? We don't need no stinkin' laws.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Daniel Morrison-2]

2008/9/16 Daniel MacKay <daniel@...>:
>> In NS, for >6 months of the year, the heat is not "waste".
>> It contributes to the overall heating for your home/office, and
>> should be subtracted from the cost of other heat sources.
>
> At Solar Nova Scotia I make this argument on a regular basis, George.
> I don't expect anyone else to get it :-)

All true, but I still feel uncomfortable recommending resistive electric
heat to anyone, when there are far more efficient ways to use electricity
for heat (i.e. heat pump), and other non-electric ways to heat (e.g.
active/passive solar).  In this sense, excess electricty converted to heat
_is_ waste, and not just in the summer months.

William Lachance said:
> Given both the soaring price of coal and the push towards even more
> expensive renewables for power generation

If the Scotian Windfields data sheets are to be believed, wind-generated
electricity is already less expensive than coal or oil.  From their '10
Myths of Renewable Energy:

#2 The current cost of wind and wave energy is higher than fossil fuels

       The cost of wind and wave energy has already been proven to be
       lower than what we currently pay for electricity in Nova Scotia.
       Not only has every contract for supply of renewable energy been
       below the residential rate (/kWh), but the costs of all renewable
       energies are decreasing, while the costs of all fossil fuels are
       increasing.

Sorry, I could only find it in PDF:

http://www.scotianwindfields.ca/Joomla/index.php?option=com_docman&task=cat_view&gid=66&Itemid=94

-D.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[D G Teed-2]

On Tue, Sep 16, 2008 at 1:07 PM, Ian Campbell <ian@...> wrote:
> You're a little late to the party in terms of the drive for lower
> power chips. The ULV Pentium M chips drew something like 5W a couple
> years ago. If you want to exclude laptop chips, VIA has been sitting
> in that niche for years. In the datacenter world look at the HE
> opteron chips.

The marketing on power reduction has been around for awhile.
But no one has produced high volumes of low power
total solutions at low cost like this.  In any case, I'm no driver, I'm
merely a user and I've been aware of the market up to now.

I don't know anyone who would trust their server room to VIA cpus,
and in any case VIA would have limitations on producing high volumes.
Sun has tried to get in on this too, with an argument on
cheapest threads per watt for their T2000.  I don't know if their
concept has actually saved any particular customer type any money,
given the cost of the T2000.  HE opterons are better than their peer
Opterons, but still eat power compared to PIII or Atom.  I'm interested
in the full power draw, as well, not just the CPU.  The chipset
and other components can be an unknown until you set up
a system and measure it.  Indeed, the heatsink and fan on the Atom
motherboard is on the chipset, while the CPU is small passive heatsink.
The numbers I gave here were measured, so that includes the inefficiency
of the power supply, fans, disks, etc.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[D G Teed-2]

On Tue, Sep 16, 2008 at 1:58 PM, Daniel Morrison <draker@...> wrote:

> If the Scotian Windfields data sheets are to be believed, wind-generated
> electricity is already less expensive than coal or oil.  From their '10
> Myths of Renewable Energy:
>
> #2 The current cost of wind and wave energy is higher than fossil fuels
>
>       The cost of wind and wave energy has already been proven to be
>       lower than what we currently pay for electricity in Nova Scotia.
>       Not only has every contract for supply of renewable energy been
>       below the residential rate (/kWh), but the costs of all renewable
>       energies are decreasing, while the costs of all fossil fuels are
>       increasing.
>
> Sorry, I could only find it in PDF:
>
> http://www.scotianwindfields.ca/Joomla/index.php?option=com_docman&task=cat_view&gid=66&Itemid=94
>

This is getting off topic, but I don't think wind, tides and sun will be able to
power our future, unless we go back to using electricity like we did in the
1940's (or perhaps 1910's).  Electricity is going to power cars, home
heating, and lots of stuff in the future (whether hydrogen cells are involved
is mute, as hydrogen will be produced by electrolysis).  We are going to need,
I would guess, double the current electricity in the coming decades,
and the only path to
doing so is nuclear.  I don't like nuclear, but I think it is the only
practical solution.

I just wish the 3 trillion pumped into the Iraq war had been put into
this instead:

http://en.wikipedia.org/wiki/DEMO

We (using the largest possible royal 'we') might have accelerated that
timeline massively.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug

[Daniel Morrison-2]

2008/9/16 D G Teed <donald.teed@...>:
> On Tue, Sep 16, 2008 at 1:58 PM, Daniel Morrison <draker@...> wrote:
>> Myths of Renewable Energy:
>> #2 The current cost of wind and wave energy is higher than fossil fuels

> This is getting off topic

Yes.

> but I don't think wind, tides and sun will be able to power our future,
> unless we go back to using electricity like we did in the 1940's (or perhaps
> 1910's).

Hmm, how to approach this...

Option 1) we won't know unless we try.  With such a sparsely populated
country, I wouldn't be surprised if Canada were able to produce all its
electricity from renewable sources (I'm including hydro electricity in this).
Don't forget also that the next generation of PV solar cells might be either
dirt cheap, or highly (75%) efficient.  If we're lucky, the generation after
that will be _both_ cheap and efficient.

Option 2) The 20th century will be remember as the time of plenty.  We WILL be
returning to usage patterns closer to those at early part of this century (per
capita)... no choice about it (except for the elite few).

> Electricity is going to power cars,

> home heating,

Man, I hope not.  Electricity is a high-grade energy; using it for heat is a
poor choice.

> and lots of stuff in the future (whether hydrogen cells are involved is
> mute, as hydrogen will be produced by electrolysis).

'moot'.

Electrolysis is a darn sight better than fossil fuel extraction, which is the
dead end being touted now.  Still, it's incredibly energy inefficient.  I've
seen proposals for direct solar water splitting.  Very slow at present, but if
worked on, could be miles more efficient than electrolysis.

> We are going to need, I would guess, double the current electricity in the
> coming decades, and the only path to

Sorry I'm being such a stick in the mud... we do not "need" that much
electricity.  We "want" that much electricity.  After the green revolution
(whether it be violent or peaceful) the attitude that we "need" relatively
unlimited supplies of cheap energy will be regarded as the backwards thinking
that got us into this mess.

> doing so is nuclear.  I don't like nuclear, but I think it is the only
> practical solution.

   "I'd put my money on the sun and solar energy. What a source of power! I
   hope we don't have to wait until oil and coal run out before we tackle
   that." - Edison to Ford and Firestone in 1931
           (from "Uncommon Friends" by J.Newton)

> I just wish the 3 trillion pumped into the Iraq war had been put into
> this instead:
> http://en.wikipedia.org/wiki/DEMO

Nevertheless, this is something I can wholeheartedly agree with.  I haven't
heard any serious arguments against fusion apart from the nay-saying "too
expensive, too hard, can't be done" sort.

Still, why spend trillions on fusion power when we have the largest, most
spectacular, and most reliable fusion power plant ever imagined just sitting
on our doorstep?

(The Sun, in case it isn't obvious...!)

-D.
_______________________________________________
nSLUG mailing list
nSLUG@...
http://nslug.ns.ca/cgi-bin/mailman/listinfo/nslug