Page 1 of 2

Power consumption, why has it got so out of control?

Posted: Mon Jun 14, 2004 12:40 pm
by Startled Pancake
As someone with a degree in microbiology Ive a vested interest in power consumption and why we all should take steps to reduce it (your mileage may vary, but mines scientifically accurate). With that in mind Ive been trying to build a more efficient, quieter PC that still performs when completing day to day tasks (including some mindless gaming). Unfortunatly Im encountering a total lack of transparency when attempting to purchase PC components based on their energy efficiency. For instance:

Official specs of Barnton CPU's:

AXMA2500FKT4C 1867 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0
AXMA2600FKT4C 2000 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0
AXMA2800FKT4C 2133 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0


All at 68 watts, ignoring the fact that power consumption/heat output scales in a linear fashion with MHz. Very helpful AMD.

The new mobile XP's appear to be a good choice but they're just Barntons undervolted to 1.45v. A Thoroughbred B (256k cache) at 1466 and 1.5v is rated at 49 watts. A mobile Barnton running at 1.45v (as little as makes no difference) and 1867Mhz is rated at 45 watts! Clearly the laws of phsyics are being bent here, even accepting the barnton may be slightly more efficient (although highly unlikely as it has 25% more transistors).

Disabling power management in desktop CPU's (A64's excepted), wheres the logic in that? Even when you can buy a decent CPU (mobile A64, P-M), you cant buy a motherboard that supports it properly.

Dont get me started on GFX cards. Im guessing you will be quickly dispatched by Nvidia's ninja hit squad if you accurately publish how much juice a Geforce 6800 Ultra sucks down. Oh and guess what, ATI despite having far more realistic wattage levels in their wisdom chose to disable power management in desktop cards...

A current PC could easily run at 99% of its current performance yet use half the amount of power, unfortunatly major hardware manufacturers have decided not to implement power saving features despite the fact that its in everyones interest for them to do so.

Even if being green isnt your bag, with a current PC gobbling up over 200 watts at full load, add the monitor (about 200 watts again for a 19" CRT), include a set of speakers, hub, ADSL modem etc and your looking at something thats costing a serious wad of cash to run over the course of its life (especially if you live in europe where your power bill can easily be over $200 a month).

I wouldnt buy a car if I didnt know what it did to the gallon, is it too much to ask how much the bits inside my PC are going to cost me on my power bill/conscience over the next couple of years? I know there have been serveral excellent posts on this forum giving some of the answers to this, but is it too much to ask for hardware manufatcurers to put accurate power consumption in their spec sheets?

/catches bus into town to buy an LCD monitor, wind generator and some sun screen...

Re: Power consumption, why has it got so out of control?

Posted: Mon Jun 14, 2004 9:25 pm
by AZBrandon
Startled Pancake wrote:As someone with a degree in microbiology
Cool, you must be really smart then.
Startled Pancake wrote:Ive a vested interest in power consumption and why we all should take steps to reduce it (your mileage may vary, but mines scientifically accurate).
Oh neat, your methods are scientifically accurate so we don't need to worry about you just making stuff up that's plainly wrong then.
Startled Pancake wrote:Unfortunatly Im encountering a total lack of transparency when attempting to purchase PC components based on their energy efficiency. For instance:

Official specs of Barnton CPU's:

AXMA2500FKT4C 1867 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0
AXMA2600FKT4C 2000 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0
AXMA2800FKT4C 2133 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0

All at 68 watts, ignoring the fact that power consumption/heat output scales in a linear fashion with MHz. Very helpful AMD.
Yep, it's curious that they claim the same power spec for all the CPU's. I think even those figures are a little on the high side too, even for the 2133mhz model, by the way.
Startled Pancake wrote:The new mobile XP's appear to be a good choice but they're just Barntons undervolted to 1.45v. A Thoroughbred B (256k cache) at 1466 and 1.5v is rated at 49 watts. A mobile Barnton running at 1.45v (as little as makes no difference) and 1867Mhz is rated at 45 watts! Clearly the laws of phsyics are being bent here, even accepting the barnton may be slightly more efficient (although highly unlikely as it has 25% more transistors).
First off, it's spelled Barton, not barnton. Second, what law of physics are you using? You already have the specs for the 1.65v Barton, and they claim 68 watts. Since you're a scientist and all, I'm sure you know the formula for calculating power consumption based on voltage, and being a scientist you arrived at the fact that dropping from 1.65v to 1.45v drops the power to 77% of the old figure. Even if we use that 68 watt figure, 77% of 68 watts is 52 watts, which is quite close to the quoted 45 watt usage.
Startled Pancake wrote:A current PC could easily run at 99% of its current performance yet use half the amount of power, unfortunatly major hardware manufacturers have decided not to implement power saving features despite the fact that its in everyones interest for them to do so.
Well, I'm not a scientist, but I do like accuracy myself, so I have a device called a Kill-A-Watt. It measures real power usage from anything you can plug in to a wall socket. Several of us on this site including the owner of the site use them and find it to be a handy device.

I have a 4-year old Dell 4100, which has a Pentium 3, 1ghz processor. It draws 61w at idle, and 81w under full load. My new system, based on AMD'd Mobile Athlon 2400+, in spite of running at 1.55v instead of 1.45 since my motherboard is an older design from before the mobile chips came out draws as little as 42w at idle, and a maximum of 83watts at full load. I'm not sure what power saving technology you're referring to, but it appears to me it uses a lot less power at idle while being about 3 times as fast at full load and drawing the same power.
Startled Pancake wrote:Even if being green isnt your bag, with a current PC gobbling up over 200 watts at full load, add the monitor (about 200 watts again for a 19" CRT), include a set of speakers, hub, ADSL modem etc and your looking at something thats costing a serious wad of cash to run over the course of its life (especially if you live in europe where your power bill can easily be over $200 a month).
As I said already, I checked my own PC and it draws as little as 42 watts, and no more than 83 even at full load. That's quote a lot less than 200 watts. I also have a 21" CRT, which is certainly bigger than a 19" and it draws an average of 115 watts, which is again, still much less than the 200 watts you claim a 19" CRT uses. Which of your "scientifically accurate" methods did you use to come up with that?
Startled Pancake wrote:I wouldnt buy a car if I didnt know what it did to the gallon, is it too much to ask how much the bits inside my PC are going to cost me on my power bill/conscience over the next couple of years?
Apples to oranges, especially in this country. Most folks have their computer on maybe 4 hours a day. Only a tiny, TINY fraction of folks leave their computers on 24x7, and even among them, very few have their monitor on 24x7. My 42w computer will thus use about 30kwh of electricity in a month. Most of the country other than the Republic of California pays 8 cents per kwh. That's $2.42 to leave your computer on 24x7. That's less than I paid for one oreo cookie shake at Jack in the Box yesterday. For that average user with their computer on 4 hours a day, they would use only about 40 cents of electricity in a month, less than I paid for a 3 Musketeers bar today.
Startled Pancake wrote:I know there have been serveral excellent posts on this forum giving some of the answers to this, but is it too much to ask for hardware manufatcurers to put accurate power consumption in their spec sheets?
Considering a computer can easily run anywhere from $600 - $2000, people just don't care if it uses 40 cents or $3 of electricity in a month. It's not like cars where you can end up going through $250 in gas if you drive a lot and get bad gas mileage. It's just not the same. If you're really interested, just buy a Kill-A-Watt yourself, stop complaining, and start doing real, legitimate testing of power draw so we can all be better informed consumers. Don't rely on someone else to do your work for you; sometimes you have to do it yourself.

Posted: Tue Jun 15, 2004 12:44 am
by nutball
Yay! Two shoulders, two chips! That's one each!

Handbags, ladies?

Posted: Tue Jun 15, 2004 1:14 am
by Bosk
I agree with Startled Pancacke completely.

The idea that the latest GeForce card will use almost as much power as a Prescott chip is just ludicrous, and is indicative of the trend in hardware for bigger/faster components but not smarter/more ergonomic ones.

But it's pretty understandable when you look at the marketing behind it. Imagine how many more gamers and overclockers will rush out and buy the latest vid card with the "20% MORE POWER" stamped on the box, rather than something like "20% more heat efficient"?
I think it's also a part of modern culture (at least American and maybe Aussie culture) that "Bigger is better", and because our world moves so fast nowdays everyone's hungering for more SPEED even in their computers just to keep up, at least that could be part of our subconscious reasoning behind these tech purchases we make.

About 2 years ago i was running an old AMD 700 with 128MB of SDRAM and a 32MB vid card, and when i finally made the choice to upgrade my rig i had no idea whatsoever that doing so would be inviting a huge amount of extra noise and heat into my livingroom, simply because i wanted a faster computer!
I've had to resort to watercooling - a pretty extreme cooling method - just to help deal with them, which i think says alot about how heat levels of modern components are spiralling way out of control.

Re: Power consumption, why has it got so out of control?

Posted: Tue Jun 15, 2004 1:19 am
by Startled Pancake
AZBrandon wrote:]
Startled Pancake wrote:As someone with a degree in microbiology
Cool, you must be really smart then.
Startled Pancake wrote:Ive a vested interest in power consumption and why we all should take steps to reduce it (your mileage may vary, but mines scientifically accurate).
Oh neat, your methods are scientifically accurate so we don't need to worry about you just making stuff up that's plainly wrong then.

This stuff is mainly immature flaming, I dont think its appropriate to the tone of these forums but Im not a moderator so Ill leave it at that. However the post itself contains so much misinformation I cant just leave it alone.
Startled Pancake wrote:Unfortunatly Im encountering a total lack of transparency when attempting to purchase PC components based on their energy efficiency. For instance:

Official specs of Barnton CPU's:

AXMA2500FKT4C 1867 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0
AXMA2600FKT4C 2000 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0
AXMA2800FKT4C 2133 MHz 266 MHz 1.65V 10 0.13 micron 41.40A 68.30W 90°C organic Barton 6A0

All at 68 watts, ignoring the fact that power consumption/heat output scales in a linear fashion with MHz. Very helpful AMD.
Yep, it's curious that they claim the same power spec for all the CPU's. I think even those figures are a little on the high side too, even for the 2133mhz model, by the way.
Startled Pancake wrote:The new mobile XP's appear to be a good choice but they're just Barntons undervolted to 1.45v. A Thoroughbred B (256k cache) at 1466 and 1.5v is rated at 49 watts. A mobile Barnton running at 1.45v (as little as makes no difference) and 1867Mhz is rated at 45 watts! Clearly the laws of phsyics are being bent here, even accepting the barnton may be slightly more efficient (although highly unlikely as it has 25% more transistors).
First off, it's spelled Barton, not barnton. Second, what law of physics are you using? You already have the specs for the 1.65v Barton, and they claim 68 watts. Since you're a scientist and all, I'm sure you know the formula for calculating power consumption based on voltage, and being a scientist you arrived at the fact that dropping from 1.65v to 1.45v drops the power to 77% of the old figure. Even if we use that 68 watt figure, 77% of 68 watts is 52 watts, which is quite close to the quoted 45 watt usage.

So Im right then, both the moblie Barton's and the desktop Barton's (my apologies) power dissapation specifications are incorrect.
Startled Pancake wrote:A current PC could easily run at 99% of its current performance yet use half the amount of power, unfortunatly major hardware manufacturers have decided not to implement power saving features despite the fact that its in everyones interest for them to do so.
Well, I'm not a scientist, but I do like accuracy myself, so I have a device called a Kill-A-Watt. It measures real power usage from anything you can plug in to a wall socket. Several of us on this site including the owner of the site use them and find it to be a handy device.

I have a 4-year old Dell 4100, which has a Pentium 3, 1ghz processor. It draws 61w at idle, and 81w under full load. My new system, based on AMD'd Mobile Athlon 2400+, in spite of running at 1.55v instead of 1.45 since my motherboard is an older design from before the mobile chips came out draws as little as 42w at idle, and a maximum of 83watts at full load. I'm not sure what power saving technology you're referring to, but it appears to me it uses a lot less power at idle while being about 3 times as fast at full load and drawing the same power.

Since you dont tell us what is contained in this PC, whats the point of this paragraph? Your whole pc consumes less power than a current high end graphics card. If you didnt know that Power Now! and Speed Step are built into all AMD and Intel desktop CPU's but disabled maybe you should have looked it up?


As I said already, I checked my own PC and it draws as little as 42 watts, and no more than 83 even at full load. That's quote a lot less than 200 watts. I also have a 21" CRT, which is certainly bigger than a 19" and it draws an average of 115 watts, which is again, still much less than the 200 watts you claim a 19" CRT uses. Which of your "scientifically accurate" methods did you use to come up with that?

Here is a resonable quality 19" CRT with a manufacturers quoted power consumption of 150W:

http://www.tigerdirect.ca/applications/ ... u=E75-1908

No doubt no brand OEM 19"CRT's will be more, although I dont have the figures to back ithat statment up (because suprise suprise, OEM's dont release those figures)
Startled Pancake wrote:I wouldnt buy a car if I didnt know what it did to the gallon, is it too much to ask how much the bits inside my PC are going to cost me on my power bill/conscience over the next couple of years?
Apples to oranges, especially in this country. Most folks have their computer on maybe 4 hours a day. Only a tiny, TINY fraction of folks leave their computers on 24x7, and even among them, very few have their monitor on 24x7. My 42w computer will thus use about 30kwh of electricity in a month. Most of the country other than the Republic of California pays 8 cents per kwh. That's $2.42 to leave your computer on 24x7. That's less than I paid for one oreo cookie shake at Jack in the Box yesterday. For that average user with their computer on 4 hours a day, they would use only about 40 cents of electricity in a month, less than I paid for a 3 Musketeers bar today.

It says specifically in my post that energy costs are significantly higher in Europe and therefore have more bearing on TCO. Seriously, your diet will kill you or at the very least make you overweight and spotty. You dont need to be a biologist to work that one out
Startled Pancake wrote:I know there have been serveral excellent posts on this forum giving some of the answers to this, but is it too much to ask for hardware manufatcurers to put accurate power consumption in their spec sheets?
Considering a computer can easily run anywhere from $600 - $2000, people just don't care if it uses 40 cents or $3 of electricity in a month. It's not like cars where you can end up going through $250 in gas if you drive a lot and get bad gas mileage. It's just not the same. If you're really interested, just buy a Kill-A-Watt yourself, stop complaining, and start doing real, legitimate testing of power draw so we can all be better informed consumers. Don't rely on someone else to do your work for you; sometimes you have to do it yourself.
The point of half my post is that I shouldnt have to buy a kill-a-watt meter but that I should be able to determine the energy effciency of my computer system by simple addition of the manufacturers specifications. Hence my request for more transparency...

If you do live in California, heres what you can expect to save buy buying an LCD monitor or any device that will save you 38 watts:

http://asia.cnet.com/itmanager/netadmin ... 480,00.htm

synopsis: 38 watts over 3 years will cost you $130. Therefore a prescott CPU will cost you nearly $250 more than a P-M. Using these figures a 200W graphics card for 4 hours a day will cost $120 over the course of its 3 year life, a significant proportion of its purchase cost. Anyone here think that energy costs are going to go down not up over the next few years?

I like many people leave a system on 24/7. I dont have any choice as these days Im a Citrix consultant (oh the unbridled joy) and I need access to a robust and failry powerful lab system at all times. If you dont then obviously proportionally power consumption is less important to you, but who are you to determine other peoples usage habits?

Youve intentionally missed the whole point of my post (why does my PC consume more power than it needs to and why cant I quickly work out what a new PC will cost to run) in order to flame and insult me, but thats your affair mate.

Posted: Tue Jun 15, 2004 1:32 am
by Startled Pancake
nutball wrote:Yay! Two shoulders, two chips! That's one each!

Handbags, ladies?
Damn your eyes sir for making me spit my mouthful of tea over my monitor screen. Funny though :)

Re: Power consumption, why has it got so out of control?

Posted: Tue Jun 15, 2004 2:48 am
by CharlieChan
Startled Pancake wrote: Youve intentionally missed the whole point of my post (why does my PC consume more power than it needs
If feel a Athlon is too powerful for you needs you can always build a PC using a C3.
and why cant I quickly work out what a new PC will cost to run) in order to flame and insult me, but thats your affair mate.
The cost of running a PC depends on how you use it - a bit like a car.

I can understand your frustration but this is probably not the place to vent your anger. If you wish to build a quiet PC (quiet=efficient is not always true) then list your requirements and I am sure many people on this forum would be able to help.

Also when you finish building your new PC have look at what these people are doing and join our team. As a microbiologist and a person who wish to help the environment you may consider helping humanity as well, after all you did mention your PC is on 24/7 :D.

Posted: Tue Jun 15, 2004 2:59 am
by trodas
Sorry, to interrupr interesting discussion, but... Anyone have clue if for Europe (you know, the rest of the world use 230V, 50Hz and not 110V 60Hz like US...) is possible to buy something nice and precise as the Kill-A-Watt seems to be...? :?: :?: :?:

Posted: Tue Jun 15, 2004 3:53 am
by CharlieChan
trodas wrote:Sorry, to interrupr interesting discussion, but... Anyone have clue if for Europe (you know, the rest of the world use 230V, 50Hz and not 110V 60Hz like US...) is possible to buy something nice and precise as the Kill-A-Watt seems to be...? :?: :?: :?:
I have one of these. Contact them to find a local reseller.

Posted: Tue Jun 15, 2004 4:46 am
by Nebor
Electricity is like gasoline, I won't care about it until it runs out. I'm not going to buy something I can't afford to run.

Posted: Tue Jun 15, 2004 6:19 am
by Rusty075
Dang....this thread has returned to civility...I missed all the excitement.

Just to add some rational thinking to all this, I'll throw in a few semi-random thoughts:

1. Power consumption data is available for nearly every PC component, if you know where and how to look. The numbers that AMD and Intel list as wattages aren't "incorrect" they're just not the numbers Pancake was looking for. Those numbers are Design Guidelines for designers of thermal solutions, not the max electrical draw of the CPU.

AMD and Intel also come about those numbers differently:

AMD has started listing the TDP as a single value for a series of similar CPU's (not only the Bartons, but the A64's as well) The number listed is the max wattage for whatever the last one of that series will be capable of. This simplifies the design solution for the mobo and HSF guys, by allowing them to design for the hottest CPU possible, with the knowledge that that same solution with therefore work for any CPU in that lineup.

Intel, on the other hand, under reports their TDP intentionally. Their logic is that it's unlikely that a user will actually run the CPU at 100% for an extended period of time, so why over-design the cooling system to allow for it. For that reason their TDP's are as much as 25% below the actual Wmax. They also have the safety net of the thermal-throttling: If your P4 gets hot enough, it'll just slow itself down anyway. Personally, I think that another motive is hiding the huge wattages that their current CPU's are capable of cranking out, to keep the marketing guys happy.

2. Max power consumption data is generally useless. People simply don't run their PC's at 100% load of every component for any real length of time, so knowing the max load isn't terribly useful for estimating electrical usage.

3. 99.44% of the market simply doesn't care.

4. There are plenty of low power options out there. check out mini-itx.com for lots of info on electrically-efficient PC's.

5. Become an educated consumer. If you don't need a 75watt VGA card...don't buy one. If you don't need a 110watt CPU...don't buy one. Vote with your wallet. And the thing is, most people are already doing that. Not because they care about electrical usage, but because the high wattage stuff is also the most expensive. Remember that the "average" PC in use today is still below 1gig in CPU speed, and is running a 16meg 2x AGP card. Probably has a total DC draw of less than 100watts max.

6. If you want to make an immediate reduction in your PC's power usage, without effecting performance at all.... get a new PSU. OEM PSU's (like you find in 95% of the PC's in the real world..the Dell's, the HP's, the Gateway's, etc) have efficiencies as low as 50% Even a $30 aftermarket PSU will likely have an efficiency of 70% or better, and the really good ones are now over 80%. (Check out the PSU section here at SPCR) If you've got a system with a 150W DC draw (that's a pretty high-end system) with a cheap OEM PSU @50% efficiency, and you swap it to a better PSU with 75% efficiency, you can save 100watts in your AC draw.

Re: Power consumption, why has it got so out of control?

Posted: Tue Jun 15, 2004 6:21 am
by aidanjm2004
Startled Pancake wrote:This stuff is mainly immature flaming, I dont think its appropriate to the tone of these forums but Im not a moderator so Ill leave it at that. However the post itself contains so much misinformation I cant just leave it alone.
I must admit, I quite enjoyed reading AZBrandon's somewhat barbed comments. His comments were indeed somewhat harsh, but I'm not sure you could reasonably argue that they actually crossed the line into "flaming" territory...

Posted: Tue Jun 15, 2004 7:12 am
by Bosk
Rusty075 wrote:6. If you want to make an immediate reduction in your PC's power usage, without effecting performance at all.... get a new PSU. OEM PSU's (like you find in 95% of the PC's in the real world..the Dell's, the HP's, the Gateway's, etc) have efficiencies as low as 50% Even a $30 aftermarket PSU will likely have an efficiency of 70% or better, and the really good ones are now over 80%. (Check out the PSU section here at SPCR) If you've got a system with a 150W DC draw (that's a pretty high-end system) with a cheap OEM PSU @50% efficiency, and you swap it to a better PSU with 75% efficiency, you can save 100watts in your AC draw.

That's a great point you make which i don't think many people consider when they're looking for a new PSU either.
I know used look for the ones with simply the highest Wattage i could afford, but now i've become smarter (thanks to SPCR!) and pay more attention to thermal design, efficiency & reliability :)

Re: Power consumption, why has it got so out of control?

Posted: Tue Jun 15, 2004 7:18 am
by jojo4u
aidanjm2004 wrote:
Startled Pancake wrote:This stuff is mainly immature flaming, I dont think its appropriate to the tone of these forums but Im not a moderator so Ill leave it at that. However the post itself contains so much misinformation I cant just leave it alone.
I must admit, I quite enjoyed reading AZBrandon's somewhat barbed comments. His comments were indeed somewhat harsh, but I'm not sure you could reasonably argue that they actually crossed the line into "flaming" territory...
I personally found his tone too harsh. AZBrandon really enjoyed making fun of his degree and his one spelling mistake. Also he did not catch Startled Pancakes point and shoot in a different direction.
Startled Pancakes examples are not 100% correct in my eyes but we should teach and not to bawl out and frighten away.

Now something about the topic: My sense is, that old 19" monitors need 100-150W and new ones 80-120W. German magazine C't tested 20 19" monitors 2 years ago or so and all are in the range of 80-120W.

You can't compare the T-Bred and the Barton. In the thread below, the T-Bred needed 10% more including PSU losses.
Mobile CPU always have lower leakage current. Desktop Barton has 7A typical and 12A maximum leakage current.
And if you take the 2033Mhz model, the calculation 68W -> 45W is correct.

T-Bred: http://forums.silentpcreview.com/viewto ... highlight=

Posted: Tue Jun 15, 2004 7:23 am
by silvervarg
Trodas:
is possible to buy something nice and precise as the Kill-A-Watt (in Europe)
There are many brands, looking fairly similar. You stick it in the wall socket and anything you want to measure in the device. All of them have an LCD and a few buttons to operate it.
Cost is usually about $40 in Sweden.

Now to the bad side. I don't think the precition is that great on them for low power usage. For a full PC that draws 100W it is quite ok to read anything from with +/-3 watts margin.
Measuring a low power device that takes ~2W average an error with +/-3 watts doesn't tell you much.
I am not sure if the more expensive devices gives much better accuracy.

Basicly they measure how many amps going through the unit and does some calculations to present on the screen. Nothing really magical about it, but still quite usefull.
I saw an instruction for a DIY, using your multimeter as an AMP meter and do some calculations on the net.

Posted: Tue Jun 15, 2004 7:31 am
by jojo4u
Rusty075 wrote:6. If you want to make an immediate reduction in your PC's power usage, without effecting performance at all.... get a new PSU. OEM PSU's (like you find in 95% of the PC's in the real world..the Dell's, the HP's, the Gateway's, etc) have efficiencies as low as 50%
Do you have some proof for this? Those big players typical adjust the size of the PSU according to the actual need. So the losses of cheap design are equal to the gains of being in the more effcient range.

Posted: Tue Jun 15, 2004 7:56 am
by Rusty075
jojo4u wrote:Do you have some proof for this? Those big players typical adjust the size of the PSU according to the actual need. So the losses of cheap design are equal to the gains of being in the more effcient range.
Actually...no, I don't have any documentation for that. :? That came from a post from our esteemed leader. :wink: I think he was refering a conversation he had with an Intel engineer. (he'll tell me if I'm wrong)

The big players, in my experience (and I've had my hands in plenty of their boxes) don't adjust the PSU's for anything except price. They set a minimum standard for wattage, and then buy whatever they can get their hands on cheap.

Besides, adjusting the "size" or rating of the PSU does little to effect the efficiency and zero effect power draw: A system that draws 150watts DC is going to pull the same wattage whether it's attached to a 200watt PSU or a 400watt. If the 200 and the 400 watt PSU's are of the same low efficiency, the total AC draw will be pretty close to the same. (look at Mike's PSU reviews, the efficiency does vary depending upon % of max PSU load, but its pretty small, 10% maybe)


RE: the flaming.

I didn't comment on it because I thought we had moved the thread past that..continuing to comment on it doesn't accomplish much.

Yes, Startled Pancake's first post came off as pompous. That's not an insult, don't take it as such. Even though unintentional, that is how it came off.

And Yes, AZ overstepped the bounds of politeness in pointing out the flaws in the SP's research. Reactionary, and unproductive.

So we're even...so move on :lol:

But I do have one question:

Startled Pancake: where did you ever come up with that name!?! LOL. I love it! :lol:

Posted: Tue Jun 15, 2004 8:27 am
by David Ruth
I've been back and forth in emails with ATI tech support regarding the power draw of the radeon 9600 series cards. Anyone who has ever gone looking on ATI or NVidia's websites will know they don't make the information readily available. Reading this thread reminded me that I was expecting a reply from them today to begin yet another iteration of them sending me links to useless information, so I went and checked my email, and lo and behold, they finally answered my question:
-------------------------------------------------------------------------------------
Regarding your request:

The information you have requested is only disclosed to our retail partners and engineers.



Regards,

Damien Da Conceicao
Customer Care
ATI Technologies, Inc.
http://www.ati.com
-------------------------------------------------------------------------------------

Gotta love it. Maybe I should try Nvidia next, though now that I know about the ninja hit squad, I am less inclined to incur their wrath.

Posted: Wed Jun 16, 2004 3:48 am
by silvervarg
The information you have requested is only disclosed to our retail partners and engineers.
Do you have the card available?
How about try out the system with a fairly well known old (low power) AGP card and measure draw with a Kill-a-watt meter. Then change to the new ATI/Nvidia card and measure same test with Kill-a-watt meter.
You should now easilly be able to calculate the power draw if you have the % efficiency of your PSU.

The AGP cards that require 2 separate power lines attached to the cards should need more than 100W. If they did not need that there would be no point in requesting the 2 separate power lines or even to add dual power connectors.

Re: Power consumption, why has it got so out of control?

Posted: Wed Jun 16, 2004 11:31 am
by AZBrandon
The messageboard software seems to freak out if I have a half-dozen different userid's quoted, so I'll do my best here. I'd like to clarify that I am not trying to "flame" anyone, however when I see someone post unverified, unsubstantiated claims that I personally know to be wrong, yet they insist that it's "scientifically accurate", that is very offensive to me. Unsubstantiated, factually inaccurate claims are the antithesis of scientific research.
AussieFellah: The idea that the latest GeForce card will use almost as much power as a Prescott chip is just ludicrous, and is indicative of the trend in hardware for bigger/faster components but not smarter/more ergonomic ones. But it's pretty understandable when you look at the marketing behind it. Imagine how many more gamers and overclockers will rush out and buy the latest vid card with the "20% MORE POWER" stamped on the box, rather than something like "20% more heat efficient"?

It is pretty crazy, isn’t it? Fortunately, many people don’t need any kind of special graphics power. I’m running my primary PC from the regular old onboard video. Since I don’t do any gaming from my PC, there’s nothing that I do which seems to require any more graphics speed than I already have. The guys that care about getting every last ounce of speed don’t care about power consumption, and most don’t care about noise either since if you’re gaming, you probably already have a headset on, or speakers going with game audio drowning out PC noise. Besides, even powerful graphics cards don’t draw nearly as much in 2D mode as they do when gaming, and I can’t imagine any die-hard gamer spending much more than a few hours a day actually gaming unless they’re unemployed or something.
Startled Pancake: The point of half my post is that I shouldnt have to buy a kill-a-watt meter but that I should be able to determine the energy effciency of my computer system by simple addition of the manufacturers specifications. Hence my request for more transparency
Yeah, it would be nice if you could get specs for every component out there, but it’s not always available. There’s a pretty good wealth of information on the internet about real power draw from CPU’s, since it varies depending on both model and usage. I’ve also personally not had any trouble looking up power specs for hard drives and CD-ROM’s, plus because I have a Kill-A-Watt. With that, I can selectively unplug things like my extra network card, CD-ROM, and so on to see how it really affects power draw. If I had a VGA card, I could compare draw between my onboard graphics and when the VGA is plugged in.

If you go back to your analogy to cars, just because the EPA says a particular car gets 25mpg city, 30 highway doesn’t mean your car will get that mileage. That’s when you take the initiative to start an Excel spreadsheet to track your gas mileage and see what your real world usage is. I tracked every single tank of gas I purchased for my first car from 0 miles to 50,000 miles. You can’t expect to have everything in the world just handed to you, sometimes you have to do the research yourself. Fortunately, because there’s lots of us all doing the same research, we can share that knowledge and increase the overall knowledge pool.
Startled Pancake: Youve intentionally missed the whole point of my post (why does my PC consume more power than it needs to and why cant I quickly work out what a new PC will cost to run) in order to flame and insult me, but thats your affair mate.
My point was that it doesn’t help for you to make wild claims about graphic cards and monitors consuming 200 watts, and refusing to do any research of your own when there’s a lot of us who quietly go about our business collecting real, factually accurate data. I know that my system can consume as little as 42 watts when outfitted for maximum efficiency. I know my 1999 vintage 21” monitor consumes 3 watts in sleep mode, 91 watts with a totally black screen, and 130 watts with a totally white screen, and about 115 watts average with my normal mix of applications displayed on screen. THAT is fact, not fiction. My point is that it’s not helpful to the pursuit of low power components when you make claims about consumption when those of us on this board already have real, actually facts on the subject.

Really, I feel that you started the whole topic on the wrong foot, right from the subject line of “Power consumption, why has it gotten so out of control?” I already mentioned that my 4+ year old Pentium 3 consumes as much or more power than my 2-month old AMD system, and the AMD system has far more processing power. Thus, reality conflicts with your theory about power consumption being out of control because in the real world, those of us who really care about power consumption actually have lower power consumption from new computers than from old ones. In fact, I even had an old Pentium 200Mhz system sitting around, believe it or not. It had RedHat 6.1 on it, so I was able to load up prime95 for testing. It draws 39 watts at idle, and 50 watts at full load. That’s from a Pentium 200!!! Again, reality conflicts with your theory, and reality always wins. Rather than coming on the board claiming power consumption is out of control, why didn’t you make a post asking if anyone has compared power consumption of old computers to new ones? I’m sure I’m not the only one to have done this.
Rusty075: 2. Max power consumption data is generally useless. People simply don't run their PC's at 100% load of every component for any real length of time, so knowing the max load isn't terribly useful for estimating electrical usage.
I agree, and my findings are consistant with that. In fact, I hate estimates in general, which is one of the reasons I got a Kill-A-Watt in the first place. Besides just showing real-time draw, it tracks the total draw and shows the time period up to 99 hours and 99 minutes. Thus, I can plug my computer in to it, and check back in 3 days to see what the total time elapsed is (say, 74 hours, 32 minutes) and the total power consumed (say, 4.07kwh). Thus, when you divide the two, you can see that the real power used was 54.6 watts per hour, on average. Further, if you wanted, you could reset the Kill-A-Watt on the first day of every month, and just check to see what your total draw was that month. In fact, if you’re like me and have an uninterruptible power supply with all your assorted computer stuff plugged in to that, you can plug your UPS into the Kill-A-Watt and then you know what all your stuff used for the month including whatever power loss there was just to provide a float charge to the UPS. Real research is work, but it’s very rewarding work for those of us that enjoy it.
jojo4u: My sense is, that old 19" monitors need 100-150W and new ones 80-120W. German magazine C't tested 20 19" monitors 2 years ago or so and all are in the range of 80-120W.
When theory and reality conflict, reality always wins. I don’t personally have any monitors available for testing besides my own, but as I already stated, my 21” monitor is from at least 1999 or 2000. I don’t know the exact age because I purchased it second hand from a company that auctions off lease return equipment and bankruptcy foreclosure seizures. Any way you slice it, my monitor is 4 to 5 years old and as I mentioned, it draws an average of 115 watts. That is still WELL below the 200 watts he claimed, plus a 21” must illuminate a larger area, and thus use more power than a 19” monitor. I am all for research and publication of findings, but I am personally offended when people willingly publish false information, such as a 19” CRT using 200 watts when there is no scientific research to back it up. If he doesn’t know what they draw, all he has to do is ask. It’s shameful to make up claims and call it “scientifically accurate” and it undermines the research done by those of us that actually own the tools to measure power usage.
Rusty075: And Yes, AZ overstepped the bounds of politeness in pointing out the flaws in the SP's research. Reactionary, and unproductive.
Personally, I am very offended when someone makes up wild claims that I know to be untrue, then tries to validate their claims by saying they are “scientifically accurate”. If Tom’s Hardware made such claims which were so easily disproved, well, I think we all know what would happen then. The fact is, publishing false information is what’s unproductive. I can publish my own personal findings about power consumption for my system if anyone is interested. Apparently this is a rather hot topic, and since I was willing to spend the money on a Kill-A-Watt, my figures are actual findings, not just estimates. Mr. Pancake already stated that he feels he should not have to buy a Kill-A-Watt in order to validate his claims, you can read it right there in his reply to my first post. I’m sorry, but I cannot allow someone to make claims that they have “scientifically accurate” information then come back and announce that they should not have to validate their claims. That is reckless and irresponsible, and more importantly it harms our community’s work to promote quieter, efficient components because that’s what factually inaccurate information does.
silvervarg: Do you have the card available?
How about try out the system with a fairly well known old (low power) AGP card and measure draw with a Kill-a-watt meter. Then change to the new ATI/Nvidia card and measure same test with Kill-a-watt meter. You should now easilly be able to calculate the power draw if you have the % efficiency of your PSU.
That’s a great idea, however in his reply to my post he said he doesn’t want to buy a Kill-A-Watt. I would test a card myself, but I don’t own a VGA card since the on-board video is plenty good enough for me, and extremely power efficient. If anyone else would like to do such research and post their findings, it would no doubt be very helpful.

Posted: Wed Jun 16, 2004 12:55 pm
by msde
Edit: Should have taken this to private messages.

I was disappointed when the initial post sparked negativity instead of discussion.

Posted: Wed Jun 16, 2004 1:57 pm
by AZBrandon
Well, you certainly don't have to buy a Kill-A-Watt, but you'll be a better informed consumer if you do. Education costs money, there's no way around it. I've purchased enough parts to build 3 computers so far this year. I did a lot of experimentation of my own to come up with a combination of parts for two computers that I kept, and I shelved the parts I didn't need. I could have sold them on ebay I suppose, but I never got around to it.

As others have pointed out, most all of the information is out there already, free of charge, for those who want to look for it. You can also enhance your knowledge if you want to spend a little more on your education and tools to reach findings of your own. The more of us that are willing to do so, the more that others will not need to.

As an example, here's a chart of the data I collected on my own system. This is power drawn at the socket, as measured with a Kill-A-Watt. The system is a Biostar iDEQ V200 running an AMD Mobile Athlon XP 2400+ at 1.55v. Additional components are a Toshiba 2.5" hard drive and Asus CD-ROM drive. The readings for "CPUidle" mean that I was running the CPUidle program while the system was idle. That's software that reduces the power consumption of your CPU to nearly zero when idle.

Image

I hope this helps someone. I can also post the usage of other various items I've measured such as my laser printer, 21" CRT, laptop, computer speakers, stereo, and so on.

Posted: Wed Jun 16, 2004 3:43 pm
by Laurent
Just to add my 2 cents:

- There are some good physical reasons why power consumption is getting out of control. When you're shrinking process:
o If you are keeping die size constant, you're adding transistors.
o You're upping frequency.
o Therefore, power goes up.
o In addition, in .13 and even more in 90, leakage is a serious issue. If you want fast transistors, leakage gets huge, so you can't afford using just fast transistors. But you need some if you want to up your frequency.

- If you look at one of the current ultra-hot processors (P4 or nVidia chips), be sure that they are very careful about power. This means that they use all kinds of power tricks to reduce power _whenever it does not impact performance_. If they didn't, we'd have 5000 watt chips by now.

- Having idle modes or clock throttling is a great thing for average power when usage varies a lot. It is not that easy to implement and takes real time and money. K8 has it (C'n'Q), but since their mobile chip is the same as the desktop chip, it is probably just by chance. For Intel, they have an incredibly good mobile processor with the whole power saving enchilada. But the P4 is an entirely different design. Thus it doesn't get "free" power saving features. As Intel has announced they are pretty much dumping the P4 architecture in favor of the Pentium M, if we're lucky, we'll get "free" power management in desktop rpocessors too.

Laurent

Posted: Wed Jun 16, 2004 4:29 pm
by CharlieChan
Hi AZ, what FSB are you running? You have fine XP-M sample, my XP-M 2400 overclock to 2162@166 1.55V measures 132W at the wall socket when running folding@home. The motherboard is a Gigabyte GA-7N400 Pro2 with only a single barracuda IV.

Posted: Wed Jun 16, 2004 4:45 pm
by AZBrandon
CharlieChan wrote:Hi AZ, what FSB are you running? You have fine XP-M sample, my XP-M 2400 overclock to 2162@166 1.55V measures 132W at the wall socket when running folding@home. The motherboard is a Gigabyte GA-7N400 Pro2 with only a single barracuda IV.
For the purpose of that test, I was running 170mhz FSB. I have since dropped it back to 166 so my minimum speed at 3x is now just 498mhz, plus now it will run stable at 12.5x166 (2075mhz), wheras before it was either 2040 (12x170) or 2125 (12.5x170), and it wouldn't do 2125 reliably. What kind of graphics card do you have? I'm running onboard graphics.

Posted: Wed Jun 16, 2004 4:51 pm
by CharlieChan
AZBrandon wrote:What kind of graphics card do you have? I'm running onboard graphics.
Passive ATI 9200 128MB, just ran CPUBURN - 141W. The motherboard does not undervolt, 1.55V is what it defaults too so I may as well overclock it.

Posted: Wed Jun 16, 2004 8:03 pm
by AZBrandon
CharlieChan wrote:Passive ATI 9200 128MB, just ran CPUBURN - 141W. The motherboard does not undervolt, 1.55V is what it defaults too so I may as well overclock it.
I'm in the same situation; no downward voltage control on my board, and it shows up as "unknown CPU" at 1.55v. Does your system also have onboard video? I'm just curious, since this would be a great opportunity to compare power usage at the plug for onboard video versus with your card installed.

Posted: Wed Jun 16, 2004 9:43 pm
by aidanjm2004
AZBrandon wrote:The readings for "CPUidle" mean that I was running the CPUidle program while the system was idle. That's software that reduces the power consumption of your CPU to nearly zero when idle.
I'm impressed by the ability of CPUidle to lower energy expenditure (as demonstrated by your rather impressive graph). You'd think any operating system worth its salt would have that kind of functionality built in.

Posted: Thu Jun 17, 2004 12:54 am
by CharlieChan
AZBrandon wrote:Does your system also have onboard video? I'm just curious, since this would be a great opportunity to compare power usage at the plug for onboard video versus with your card installed.
No. It is a Nforce2 ultra 400 motherboard with onboard RAID, SATA and Firewire. Both RAID and SATA are disable at the BIOS to speed up booting.

Posted: Thu Jun 17, 2004 2:53 am
by NetTechie
Not to hijack the topic, but has anyone stacked the Kill A Watt to see how much power it itself consumes? :lol:

I'm excited to hear there is such a thing as a kill-a-watt (I'm a noob to this stuff). Lookin to buy this one on ebay.

I'd like to mention that the "tiny fraction" of people that leave there computers on 24/7 might amount to every corporation in America. I know that a callcenter I worked for had 145 computers in it that were on at all times. I know an isp I worked for had about 30+ servers on at all times. I know a computer repair shop I worked for had 5 systems on at all times. In fact.... tryin to think... don't know any business that shut them off at night now that I think of it! My point, if you own a business this could be a very serious issue, effecting your electric bill and monthly $ output. I leave my computer on about all the time, and all my gaming friends do also (sometimes they shut them down... but it's rare). They stay in the teamspeak server and online in chat, so this is the way I know this. I really don't think it's a small fraction that leave them on all the time... in fact, I think it's a fairly large amount. Probly the majority of the PC's in existence in the world are on at this time, being businesses have the majority I'd bet. If they (hardware makers) really took into account power consumption it might have a rather noticeable effect on power consumption in large commercial cities. In fact.... I think in a PSU review on this site I read something about Japan(?) having very strict "efficiency" guidelines for all electrical devices. Now... something like that could really bring a revolution in hardware if it were made an issue in USA.... ofcourse that'd probly drive hardware prices up, so I don't know if that would be so good. :D Just my two cents on power consumption, and the "TINY fraction" of pc users who leave computers on 24/7.

Lookin forward to getting a kill-a-watt.... see if that refrigerator is really what's driving up the electric bill lately. 8)