PC processor power consumption

  • Thread starter Thread starter U. U.
  • Start date Start date
U

U. U.

I am going to finally replace my 6 year old IBM Aptiva. Since I expect the
cost of electricity to go up steeply after the election, my main goal is to
keep down power consumption and heat.

I know some of the AMD processors can run at only about 5 watts in standby
mode, but I can't find similar info on the Intel processors. Is there some
website (or two) where I can compare the heat and power consumption of
processors? And how do I know if a processor can "hibernate" (copy RAM to
HD to shut down the memory chips) and which chip sets support that?

Thanks for any help you can give me.

Ule
 
I am going to finally replace my 6 year old IBM Aptiva. Since I expect the
cost of electricity to go up steeply after the election, my main goal is to
keep down power consumption and heat.

A few points to remember:

The most power hungry processors working on their most power-intensive
code top out at about 100W. A typical CRT monitor consumes 100-150W
all the time that it is turned on. If you're looking at reducing
power consumption of your computer, don't even bother thinking about
anything else before buy an LCD monitor to replace that CRT one.

Also of note, most processors consume a max of about 60W under load,
this is the same as standard incandescent light-bulb consumes. Of
course, that's nothing as compared to your stove (1000W+ for the
burners and much more for the oven), microwave (800W or 1000W are
normal), refrigerator, etc. Heating and cooling are really where your
power consumption goes. Most computers only make up a fairly small
percentage of your electrical bill.

One final point, within computers you should never underestimate the
efficiency of the power supply. Most low-cost/low quality power
supplies are only about 55-60% efficient, so a chip consuming 60W
internally could easily be pulling 100W from the wall. However many
good power supplies are available with efficiency ratings up in the
95%+ range.
I know some of the AMD processors can run at only about 5 watts in standby
mode, but I can't find similar info on the Intel processors. Is there some
website (or two) where I can compare the heat and power consumption of
processors?

www.sandpile.org has a lot of good info, though I don't think it lists
the standby power consumption of any chips. For more detailed info,
www.intel.com is the place to go. Intel has the best documentation
for their processors of any company in the industry (AMD is the only
other company that comes anywhere close). Their datasheets should be
easy enough to find, though they aren't always the easiest read.
Also, when reading these comparisons, always keep an eye on just how
power consumption is being measured, often comparing the numbers from
two different companies is an exercise in futility due to vast
differences in how they are measuring.
And how do I know if a processor can "hibernate" (copy RAM to
HD to shut down the memory chips) and which chip sets support that?

This is almost entirely a function of the motherboard, and as a
generally rule it's implemented VERY poorly. I've found that only
about 10% of current boards can do a suspend-to-RAM properly. While
this is up from about 1% of a few years ago, it's still not very good.
 
One final point, within computers you should never underestimate the
efficiency of the power supply. Most low-cost/low quality power
supplies are only about 55-60% efficient, so a chip consuming 60W
internally could easily be pulling 100W from the wall. However many
good power supplies are available with efficiency ratings up in the
95%+ range.

Sorry, but I can't find any. Can you name one?

All these "Max-Performance ATX with PFC" PSUs are spec'ed for 70%
efficiency:
http://www.pcpowercooling.com/products/power_supplies/maxperformance/turbocools/index_hp_atx.htm

Zalman ZM400B-APS:
http://www.zalman.co.kr/usa/product/zm400b-aps.htm
"Designed with a high switching frequency and low power-loss
circuitry, the efficiency of this product exceeds 75% (at full load)."

Antec Power Supplies
Model: TrueControl 550PEC & TrueControl 550PGB
http://www.antec.com/specs/truecontrol550_spe_EU.html
Efficiency > 68%

Antec Power Supplies
Model: True550 EPS12V
http://www.antec.com/specs/true550EPS12V_spe.html
Efficiency > 68%


- Franc Zabkar
 
U. U. said:
I know some of the AMD processors can run at only about 5 watts in
standby mode, but I can't find similar info on the Intel processors.
Is there some website (or two) where I can compare the heat and power
consumption of processors? And how do I know if a processor can
"hibernate" (copy RAM to HD to shut down the memory chips) and which
chip sets support that?

You can take a look at Chris Hare's page:
http://users.erols.com/chare/elec.htm
 
If you're looking at reducing
power consumption of your computer, don't even bother thinking about
anything else before buy an LCD monitor to replace that CRT one.

Unfortunately, the likelihood of the cost savings in energy paying for the
difference between the two technologies is almost nill. You'd save much
more money buying the CRT and enabling power saving in the monitor.
Most low-cost/low quality power
supplies are only about 55-60% efficient, so a chip consuming 60W
internally could easily be pulling 100W from the wall. However many
good power supplies are available with efficiency ratings up in the
95%+ range.
^^^^

I highly doubt it. We use some single and dual output switchers in the
upper 70s to mid 80s, (We use 'em in embedded applications) but I've never
seen any AT/ATX supplies anywhere near the 95+% range. Even a high quality
DC/DC would have difficulty reaching that figure.
 
Unfortunately, the likelihood of the cost savings in energy paying for the
difference between the two technologies is almost nill. You'd save much
more money buying the CRT and enabling power saving in the monitor.

^^^^

I highly doubt it. We use some single and dual output switchers in the
upper 70s to mid 80s, (We use 'em in embedded applications) but I've never
seen any AT/ATX supplies anywhere near the 95+% range. Even a high quality
DC/DC would have difficulty reaching that figure.

Actually, it's quite easy to see why such high efficiency figures are
unachievable. AFAICS, most of the power loss would occur in the
Schottky diodes on the secondary side of the switching transformer.

Here is a typical 200W ATX P/S:
http://www.pavouk.comp.cz/hw/en_atxps.html

The +5V diode is an ESAD83-004 (30V, 40A), the +12V is a CTX128, and
the +3.3V is an FSF10P04B (400V, 10A).

Some datasheets are here:
http://www.scut-co.com/maindoc/techtrade/fuji/documents/diode/SBD/ESAD83-004.pdf
http://www.niec.co.jp/engver2/seihin/engpdf/_fsf10a40b.pdf

I don't have the specs for the above DTK PTP-2038 SMPS, but if we
assume that the current loads are 25A @ +5V, 8A @ 3.3V, and 4A @ 12V
(199.4W total), then the power dissipation in the diodes is at least
0.6 x 37 = 22W. This amounts to an efficiency of 90% (200/222), and
this is without considering the losses in the fan (2W?) and the
chopper transistor(s).


- Franc Zabkar
 
Tony Hill said:
within computers you should never underestimate the efficiency
of the power supply. Most low-cost/low quality power supplies
are only about 55-60% efficient,
However many good power supplies are available with efficiency
ratings up in the 95%+ range.

Almost all PC power supplies for desktops are only 65-70% efficient,
with the highest advertised efficiency being around 80%. The best
efficiency I've ever seen for a switching supply of any type was 96%,
but that's rare, and 85-90% is much more common (the voltage
regulators found on motherboards are often 90%). I'd like to know
where you got your numbers.
 
Unfortunately, the likelihood of the cost savings in energy paying for the
difference between the two technologies is almost nill. You'd save much
more money buying the CRT and enabling power saving in the monitor.

Yup, same goes for many other components in a system.
^^^^

I highly doubt it. We use some single and dual output switchers in the
upper 70s to mid 80s, (We use 'em in embedded applications) but I've never
seen any AT/ATX supplies anywhere near the 95+% range. Even a high quality
DC/DC would have difficulty reaching that figure.

I stand corrected. I think I was reading a bit too much into the
marketeering of some power supply companies, their "95%" figure was
probably some sort of "95% of the ideal efficiency if you don't count
these 10 areas of loss" or some such nonsense. I did a bit more
looking and indeed most are down in the 65-70% range for the best
ones. Still, this is a noticeable improvement over the el-cheapo
power supplies and will probably make as much or more of a difference
than any change in processor.
 
I stand corrected. I think I was reading a bit too much into the
marketeering of some power supply companies, their "95%" figure was
probably some sort of "95% of the ideal efficiency if you don't count
these 10 areas of loss" or some such nonsense. I did a bit more
looking and indeed most are down in the 65-70% range for the best
ones. Still, this is a noticeable improvement over the el-cheapo
power supplies and will probably make as much or more of a difference
than any change in processor.

Actually, I think what you were reading was probably the usual bluff
by some manufacturers about Active PFC. I noticed that many of them
claim that their >95% PF PSU will save you tons more money because
it's more "efficient", thus leading to your misconception.

As for the 65~70% figure, I think that's their full load figure, at
half load, which most of the good >300W PSU are usually at, it's
probably closer to 60%, which incidentally is the latest ATX
requirement for light load efficiency. So I'm guessin the existing
ones on market either just make the grade if at all. These are mostly
consumer grade stuff after all :P

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code
 
Yup, same goes for many other components in a system.


I stand corrected. I think I was reading a bit too much into the
marketeering of some power supply companies, their "95%" figure was
probably some sort of "95% of the ideal efficiency if you don't count
these 10 areas of loss" or some such nonsense. I did a bit more
looking and indeed most are down in the 65-70% range for the best
ones. Still, this is a noticeable improvement over the el-cheapo
power supplies and will probably make as much or more of a difference
than any change in processor.

I doubt that those "el-cheapo" PSUs have any worse efficiency figures
that the more expensive ones. In fact I see no evidence that your
statement is anything other than a wishful inference based solely on a
price tag.


- Franc Zabkar
 
Actually, I think what you were reading was probably the usual bluff
by some manufacturers about Active PFC. I noticed that many of them
claim that their >95% PF PSU will save you tons more money because
it's more "efficient", thus leading to your misconception.

AFAICS, a 99% PF PSU will save you not a single cent over a 65% one.
This is because the *real* power dissipated by either PSU is the same,
and it is this real power that is metered by the electricity supplier.
At least that's how it works for domestic loads in my locale
(Australia).
As for the 65~70% figure, I think that's their full load figure, at
half load, which most of the good >300W PSU are usually at, it's
probably closer to 60%, which incidentally is the latest ATX
requirement for light load efficiency. So I'm guessin the existing
ones on market either just make the grade if at all. These are mostly
consumer grade stuff after all :P


- Franc Zabkar
 
U. U. said:
I am going to finally replace my 6 year old IBM Aptiva. Since I expect the
cost of electricity to go up steeply after the election, my main goal is to
keep down power consumption and heat.

I know some of the AMD processors can run at only about 5 watts in standby
mode, but I can't find similar info on the Intel processors. Is there some
website (or two) where I can compare the heat and power consumption of
processors? And how do I know if a processor can "hibernate" (copy RAM to
HD to shut down the memory chips) and which chip sets support that?

I wonder if one of those "new fangled" active PFC power supplies wouldn't give
you more benefit (?).

AJ
 
Back
Top