Power Usage revisited, a case study.

[stanger89]

This topic comes up from time to time, is it better to leave your PC on all the time, or turn it off. Well, I'm not even going to attempt to tackle that, that's far to complicated, there's far too little good info about it's consequences (specifically HDD spinning up/down), but most of all, that's a personal decision.

However, I've recently had cause/motivation to look into my energy usage. I moved into my house almost a year ago (11 months) and since then I've configured my great SageTV setup:

ReadyNAS X6: 4x 500GB (1.5TB)
"Thunderbird" SageTV server, Athlon XP 1800+, 8x 250GB, 300GB, 200GB, 160GB
"Manchester" desktop/encoding server, Athlon X2 4200+, Geforce 6600GT, WD Raptor 74GB.
"Newcastle" SageClient/HTPC, Athlon 64 3400+, Geforce 6800, Toshiba 40GB (laptop drive) - old
"Brisbane" SageClient/HTPC/Blu-ray player, Athlon X2 BE-2400 (2.3GHz dual-core), Gigabyte AMD 780G motherboard, Toshiba 40GB, 2GB Corsair XMS
SageTV Extender

I'm very happy with the functionality of the setup, and due to remodeling and a late move-in to my house electricity was no issue as I had built up a nice credit over the first 6 months I owned the house and payed no electric bill over the second six .

But now I'm paying again, and after looking at my bill, I wanted to know exactly where my electricity was going. So, Kill-a-Watt in hand, I set off to determine what power was going where. I'm not quite done yet, I need another probably two-three days to finish my measurements, and I'll update as appropriate but I thought the current results might hold interest for my fellow Sage enthusiasts.

For the past couple days I've been measuring my various SageTV devices over the span of ~1day with the Kill-a-Watt. For those unfamiliar with the device, you plug it into the wall, and plug your appliances into it. It has an LCD screen and will report the current wall voltage, frequency, current (Amps), power in Watts and VA (and you're either and EE or a true geek if you know or care what VA is ), Power Factor, but what's really useful, is it gives you KWH used and the time over which it's used.

The later is important because while you can figure KWH from Watts, Watts is an instantaneous measurement, KWH is a cumulative one, that takes into account both "peak" and "off peak" usage (ie when you're using the device vs it just being idle.

I suppose that's enough verbiage for the intro, you all probably want to see the numbers don't you:

ReadyNAS X6  70W   1.66KWH/day
Thunderbird  197W  4.76KWH/day
Manchester   157W  3.77KWH/day
Manchester   128W  3.06KWH/day (Cool 'n Quiet enabled)
Newcastle    115W  2.76KWH/day
Brisbane     45W   1.08KWH/day* (Cool 'n Quiet enabled)

*This is predicted usage, since I turn off or S3 the HTPC.

Now, that's all very interesting but doesn't really mean much in and of itself. It is, perhaps, more useful, to look at it over the span of 30 days, a typical billing cycle.

Knowing you usage is interesting, but you need to have some perspective for it to be truely meaningful. So how does that relate to total energy usage. Keeping things relatively simple, average total usage (on the bill) is 800KWH/month, or over about 30day.

ReadyNAS X6  70W   50KWH/30days
Thunderbird  197W  142KWH/30days
Manchester   157W  113KWH/30days
Manchester   128W  93KWH/30days (Cool 'n Quiet enabled)
Newcastle    115W  83KWH/30days
Brisbane     45W   32KWH/30days (Cool 'n Quiet enabled)

Some quick math and you'll realize that what I have enumerated above accounts for about half of my electric bill. That is when I really started getting interested. So how much does all this cost?

Well, my average cost/KWH is $0.112, a little more math and:

ReadyNAS X6  70W   $5.60/30days
Thunderbird  197W  $15.96/30days
Manchester   157W  $12.65/30days
Manchester   128W  $10.37/30days (Cool 'n Quiet enabled)
Newcastle    115W  $9.26/30days
Brisbane     45W   $3.63/30days (Cool 'n Quiet enabled)

I'm not really sure what I'm going to do with this info. But there are a couple things that are clear, a NAS is pretty cheap/efficient to run. One thing that's interesting is that there doesn't seem to be a correlation between HDDs and usage, and if you look at the data sheets, that's to be expected, HDDs only pull about 8W spinning.

-edit

OK, now that I've got measurements for all my Sage-involved PCs (which happen to be all my PCs ), I think it's time to look at the "big picture".

Looking at my usage history since I moved in, I've had a low usage of 771KWH, and a high of 944KWH, with an average of right about 850KWH. Going through the numbers I've accumulated so far, my entire PC/Sage setup pulls about 370KWH/month, or 43% of my electric usage.

Hm, gonna need to look into that.

-edit

Added new Brisbane/780G based system.

[hemicuda]

my UPS reports about a 600VA (I'm an EE ) load most of the time through apcupsd (sourceforge). power here is about $0.0725/kWh. adds up to between 1/3 & 1/2 of my monthly bill according to the calculator program I have. That covers the system in my sig, the network gear, an HP Vectra that I'm using as a firewall/router, and 2 monitors. I power them off after about an hour of idle though. The LCD in the family room is rated about 245W per the manual. This laptop has a 130w adapter, but I don't know exactly what it draws.

[paulbeers]

For this very same reason, I turn off my HD clients when not in use (especially the one hooked to my big screen as that is usually only used for movies). It just absolutely killing my electric bill (or has to be anyway). Okay now I am going to have to buy one. Just one more gaget for me to buy.

As a side note, to help circumvent my rising energy bill, I have switched any incandecent lightbulb in my house to a compact flourencent where I can (any that are on dimmers or any can lights still have the old bulbs). This has helped my energy bill quite a bit (especially in the basement laundry room where my wife tends to forget to shut the lights off after she is in there and we realise is 2 days later!).

[stanger89]

Originally Posted by paulbeers

Just one more gaget for me to buy.

Gadget's are cool, viv la gadgets!

At the very least I think I'm going to have to investigate rebuilding my server to be optimal power/efficiency wise. I've kind of wanted to do it for a while. At least now I'll be able to guestimate how long it would take me to pay for such a system.

Also interesting, though unfortunately over a year old:
http://www.silentpcreview.com/article313-page1.html

[trini0]

Nice post.
I know how you feel concerning electricity usage.
This is one of the reasons, why I recently converted all of my network servers into Virtual Machines. So I went from 4 physical boxes (with plans to build another 2) to 1.
I'm in the planning stages to build a Sage Server, and a storage appliance.
Come hell or high water, I do not want this Sage server running 24/7, if its not being used by a client.
Maybe with everyone squeezing the dollar more and more these days, people may actually care about polls like this, where Sage can implement additional power management features within the software.
Maybe now that you put actual figures to the problem, change can be ushered in...

[stanger89]

Originally Posted by trini0

Nice post.
I know how you feel concerning electricity usage.

Actually I haven't made up my mind yet, I'm not sure if I care or not, or to put it in a more "politically correct" phrasing, I'm not sure if it's economical to "fix" my current energy usage.

Consider, my Sage server costs me $194/yr to run. With the "difficulties" of spooling up 10 HDDs at a time, I've pondered transitioning the ~2TB or so in it to a new NAS with fewer larger drives. Judging by my current NAS if I could replace my Sage server with another NAS, it would save me about $120/yr.

Sounds great, lets do it. Problem is it would cost me at least $1700 to do that, so it would take me over 14 years to make up the investment. Plus that ignores the power usage of the server I'd still need.

Come hell or high water, I do not want this Sage server running 24/7, if its not being used by a client.

I'm actually definitely leaning toward the server not being the problem. Plus spooling up numerous HDDs is a tall order and not something I want to put a PSU through unnecessarilly. And then there are the issues of the clients reconnecting to it if it's in standby. All told I'm probably burning ~40W in HDDs in my server, that means the rest of the system must be burning the rest, or about 160W.

That seems awfully high, my new Athlon X2 rig, complete with 6600GT only burns 130W with the LCDs off. According to the SCPR article, I should be able to get a system down to below 50W (excluding HDDs).

Interesting calculation:
100W run 24/7 equals 72KWHs over 30 days, or for me about $8/month.

Maybe with everyone squeezing the dollar more and more these days, people may actually care about polls like this, where Sage can implement additional power management features within the software.

You do realize you can configure the System to got to S3 or Hibernation when Sage isn't recording right?

Maybe now that you put actual figures to the problem, change can be ushered in...

I don't think you're going to get a lot of support to completely shut down the PC, especially when you can get all the energy-saving benefits, with none of the missed-recording problems with the current power management support.

[trini0]

Originally Posted by stanger89

You do realize you can configure the System to got to S3 or Hibernation when Sage isn't recording right?


I don't think you're going to get a lot of support to completely shut down the PC, especially when you can get all the energy-saving benefits, with none of the missed-recording problems with the current power management support.

Actually not all the time.
Long Story Short, once an XP PC is in a domain with users with limited accounts, power management goes out the window.
Apparently by some bug or oversight, and is supposed to be fixed in Vista.
This affected me when I was using Sage on a stand alone PC.
I tested the domain scenario a couple of weeks ago, with Sage running as a server, and s3 power management (sleeping/waking) seems to work there, but Im not going to log into it with a regular user account.
I do have an idea about power management with respect to "PC" clients (not extenders), but I will work on that when I get the server operational.

Anyway, just my 2cents. It would be interesting to see where this post goes...

[mayamaniac]

With 11 harddrives in my server (3x500G, 3x200G, 5x250G) plus a DVD, I've been monitoring power consumption also. And there's still room for one more drive in the case which means another 500G added soon. Anyway, this is an interesting topic or concern for me as well.

Running a server, of course it needs to stay on 24/7, so shutting it down is not an option. This is not just a SageTV server, it is a File Server.

Stanger89, have you tested power usage between having the harddisk powered down when idling versus having it spinning all the time? Does a NAS device have such a feature? I'm not sure if it affects power usage or increases the life of a drive or not.

I do have a UPS backup, so I'll have to monitor its power usage from there, never thought of that till now. Thanks for the info, hemicuda.

[stanger89]

Originally Posted by mayamaniac

Stanger89, have you tested power usage between having the harddisk powered down when idling versus having it spinning all the time?

No I haven't.

Does a NAS device have such a feature? I'm not sure if it affects power usage or increases the life of a drive or not.

The ReadyNAS does have the capability, and I did try enabling it, but I set a rather long timeout, 3hrs IIRC, and I don't think it ever spins them down.

[elaw]

Originally Posted by trini0

Long Story Short, once an XP PC is in a domain with users with limited accounts, power management goes out the window.
Apparently by some bug or oversight, and is supposed to be fixed in Vista.

Trini0,

Do you have a link that has more specifics about this? This is something I've been fighting with (mostly unsuccessfully) for years, it would be nice to find out more about it!

One other data point: my workstation at home used to go in/out of suspend very cleanly, then after a bunch of Windows updates began having problems. A few days ago, I uninstalled Microsoft Update (went back to Windows Update) and suspend seems to be working again!

[elaw]

Originally Posted by mayamaniac

Stanger89, have you tested power usage between having the harddisk powered down when idling versus having it spinning all the time?

I'm not Stanger, but...
I've got a kill-a-watt too (great device, by the way, and cheap! about $35) and have done some research on this. I built up a test machine and didn't try spinning down the drives but instead disconnected them.

My test system (an old PIII) drew about 70W with the hard drive disconnected. With the drive connected, during spinup it would draw 90-95W. With the drive idle (spinning but no I/O), it would draw 75W. With the drive actively seeking (machine booting up), power draw was around 85 watts.

I ran this test with several different drives, got very similar results with all. The drives were all 7200RPM IDE, in the 20-80 gig range.

So it looks like when a HD is just "sitting and spinning" it doesn't draw much, but the draw increases considerably as soon as you ask the drive to do anything.

[trini0]

Originally Posted by elaw

Trini0,

Do you have a link that has more specifics about this? This is something I've been fighting with (mostly unsuccessfully) for years, it would be nice to find out more about it!

One other data point: my workstation at home used to go in/out of suspend very cleanly, then after a bunch of Windows updates began having problems. A few days ago, I uninstalled Microsoft Update (went back to Windows Update) and suspend seems to be working again!

Its been a while, but let me see if I can unearth that info again....

[stanger89]

Originally Posted by elaw

So it looks like when a HD is just "sitting and spinning" it doesn't draw much, but the draw increases considerably as soon as you ask the drive to do anything.

I wonder how much of that extra draw is due to what the PC is doing to get the drives to read/write.

According to WD's electrical specs, 250GB drives draw ~8W spinning idle, and just a couple 1/10th of a watt more when "active".

[stanger89]

First post is updated with real numbers from "Manchester": Athlon X2 4200+, Geforce 6600GT, 2 LCD monitors (off after 5 min), Cool 'n Quiet not enabled.

[TakeFlight]

Originally Posted by stanger89

Keeping things relatively simple, average total usage (on the bill) is 800KWH/month, or over about 30day.

ReadyNAS X6  70W   50KWH/30days
Thunderbird  197W  142KWH/30days
Manchester   157W  113KWH/30days
Newcastle    TBD   TBD

Some quick math and you'll realize that what I have enumerated above accounts for about half of my electric bill.

If that's true (a few PCs being half your electric bill), I want your electric bill instead of mine. A PC running 24/7 is only a minor blip on my electric bill. And I actually don't have that high an electric bill compared to what I've heard that some other people I know pay. But, by far the things that make my electric bill what it is are my heat pump/AC, electric dryer, fridge, stove/oven, etc. Most of those things consume many hundreds of watts (in some cases thousands). So, a 150-200W PC is a blip for sure.

Although, I have done my part to conserve my PC electric usage by combining my SageTV server and my HomeSeer server into one box running 24/7 instead of two boxes running 24/7.

You have made me interested in what my various SageTV clients and server use so I'm going to borrow a Kill A Watt we have at work to find out.

[stanger89]

Originally Posted by TakeFlight

If that's true (a few PCs being half your electric bill), I want your electric bill intead of mine. A PC running 24/7 is only a minor blip on my electric bill.

That's what I thought too, but then when I looked at my usage history, and saw no drop after August, like one would expect after being able to turn off the air, I decided it was time to investigate.

And I actually don't have that high an electric bill compared to what I've heard that some other people I know pay.

Yeah, I've heard some horror stories, but it's not really the magnitude that's caught me, it's the proportion.

But, by far the things that make my electric bill what it is are my heat pump/AC, electric dryer, fridge, stove/oven, etc. Most of those things consume many hundreds of watts (in some cases thousands). So, a 150-200W PC is a blip for sure.

True, but if you think about it, most of those only draw those high powers for short periods of time. Especially things like the dryer/oven/stove, those are probably used only a couple times a week to maybe once/twice a day.

The fridge (if it's decent) probably only runs a small percentage of the time. The Air is the one that really shocked me, I ran my air last summer, but there wasn't an associated blip on my electric bill like I'd have expected.

[TakeFlight]

Originally Posted by stanger89

That's what I thought too, but then when I looked at my usage history, and saw no drop after August, like one would expect after being able to turn off the air, I decided it was time to investigate.

I definitely see seasonal patterns in my electric bill. My house is all electric (no gas/oil/etc.). My heating/cooling is from a heat pump which is a top of the line Carrier Infinity system installed 2 years ago. My electric usage can literally triple in months like Jan. and Feb. compared to months like May and Oct. In the summer Jul. and Aug. are close to double the May/Oct. numbers. So, the peak winter months cost me more than the peak summer months which in turn cost me more than the lowest usage months in the fall/spring (when you would expect them). The point being, that clearly my heat pump (and it's a new, very efficient model) has the biggest single impact on my electric bill. Even during those lowest usage months, I'd venture to guess that my heat pump still accounts for half my electric bill. It's easily 70-80% of my bill in Jan/Feb. So, that's why I don't get too concerned about how much my SageTV server is using because even if I eliminated it, my electric bill isn't going to go down by a noticeable margin. Although, what is "noticeable" is different to each person.

Originally Posted by stanger89

True, but if you think about it, most of those only draw those high powers for short periods of time. Especially things like the dryer/oven/stove, those are probably used only a couple times a week to maybe once/twice a day.

No question about it. But when you combine their usage over a month it still probably blows away a few PCs running 24/7. Of course, it does depend on how much you use those appliances, which can vary greatly from house to house.

Originally Posted by stanger89

The Air is the one that really shocked me, I ran my air last summer, but there wasn't an associated blip on my electric bill like I'd have expected.

Again, I'd like to have your electric bill if you get no blip by running your air!

[hemicuda]

TF,

check your wall thermo next time the heat pump comes on on a 'cold' day. mine would typically run for 30-45sec then kick in the "Aux" heat (could hear the relay click in). That dang thing is a major resistive element like that in a dryer. Heck, if I adjusted the temp up more than 1 degree above ambient the thing would kick in. Found the manual on the unit and unhooked the wire controlling that element. Made over a $30 difference in power usage and we didn't notice any difference in the heat. YMMV however.

[stevech]

The HVAC and refigerator(s) dominate my power bill.
And the 250w TV set that's on like 6 hrs a day. My office PC pulls 35W and the LCD monitor is more yet, according to my VOM, and it's on 4+ hrs a day.

The PC (laptop, frugal) that runs sage is like 10W.

I'd not agree that the PC is a significant portion of my power bill.

I did do a spreadsheet on this some time ago and it supports this assertion.

[TakeFlight]

Originally Posted by hemicuda

TF, check your wall thermo next time the heat pump comes on on a 'cold' day. mine would typically run for 30-45sec then kick in the "Aux" heat (could hear the relay click in). That dang thing is a major resistive element like that in a dryer. Heck, if I adjusted the temp up more than 1 degree above ambient the thing would kick in. Found the manual on the unit and unhooked the wire controlling that element. Made over a $30 difference in power usage and we didn't notice any difference in the heat. YMMV however.

Thanks for the info. However, I did do extensive research on heat pumps a little over 2 years ago before I had my new heat pump installed. I got the high end Carrier Infinity unit because it's a continuously variable speed blower along with 2 speed compressor as well as 3 stage aux/backup heat. So, the idea is this heat pump only uses what it needs to use to maintain the temp. So, most of the time it's running on low speed across the board. However, on really cold days (where the outside temp doesn't get out of the teens or twenties) just about any heat pump will need some aux/backup heat to maintain the indoor set point unless you like to keep your house on the cool side. I keep my house at 72 in the winter, which many people seem to think is high, but I like it warm and comfortable and I'm willing to pay for it. The reason my bills are 3 times as high (actually it's more like 2.5x) in the winter compared to the fall is simply the fact that the heat pump is easily running 3 times as much during those peak months. So, I don't think the aux/backup heat is really an issue with my new heat pump. Now, the old heat pump that it replaced, well that was a different story.

Sorry to the OP if this is slightly hijacking this thread. We don't need to discuss heat pumps anymore.

Back on topic. I measured my SageTV/HomeSeer server tonight with a Kill A Watt and all the connected equipment that runs 24/7 in my basement (this includes security cameras, an HDHomeRunm cable modem, switches, wireless routers, etc.) and the average wattage reading bounced around 225-228W. Not too bad considering that's a computer with 4 hard drives as well as 11 other devices/components combined.

The real shocker (in a good way) to me was the wattage of my Fujitsu Stylistic LT C-500 tablet PC that I have on my kitchen counter running 24/7 as a HomeSeer remote control. This tablet PC (which is a P3 500MHz) along with it's screen backlight on and hard drive spinning only consumes about 20-25W! I thought for sure it would be more like 40-50W. So, that was a pleasant surprise. Of course, not that it would make much difference on the electric bill either way.