I've just spent a few minutes measuring the power consumption of some of my IT equipment.  I don't have anything fancy like a Kill a watt (guess I should see whether I can get one which works on 240V/60Hz), but I'm just using the figures from my APC UPS (reads to the nearest 4W).  I'm sure that others have reported on similar experiments, but I wanted to add my findings.

 

At idle (no disks spinning) my unRAID server uses about 60W.  For each spinning disk, the consumption increases around 4 - 5 watts, so with all seven disks spinning, I get a reading of 92W.  My 16 port Gb switch (TP-Link) adds 8W to the total load.

 

This is all okay, and as good as I might have expected.

 

Now, here is the frightening info ... adding my Desktop Ubuntu machine and LED monitor increases the consumption by another 100 watts.  Until very recently, I was leaving this machine on 24/7 because it is the platform I was using for torrent and other downloads.  Now I have installed the Transmission plugin on the server, I can turn that machine off sometimes - but even then, just plugged in and with the socket turned on, it still uses more than 20W - equivalent to running a couple of energy saving light bulbs.

 

I have now resolved to always turn the computers off at the power socket when not in use.

 

This also increases my urgency for finding a headless equivalent  for Jdownloader, so that I can run that on my unRAID server (or else I have to go virtual on the server, and install an Ubuntu VM).

 

I still want to experiment with my adsl router (which is on a separate, non-smart UPS) - judging from the weight, I suspect that the power brick is a linear device, and I might be able to make some saving by replacing it with a SMPSU.

 

Edit:Edit:

Adding the router to the UPS seems to increase the load by 4W - not enough to worry about.

Go esxi... Its actually quite easy to setup and as far as I can see quickly in your specs you should be ok with it.. I have and power consumption has actually allmost not changed from having the box run only unraid to now running  an unraid VM and an Media VM that does sabnzb, transmission, sickbeard, couchpotato and maraschino.

 

I controll all from my pc that goes into full standby after not beiing used for an hour and from my iphone.

Got my machine down to 40W idle with no discs spinning.

 

If I could have a temperature controlled fan setup to turn them off <40C I could shave another 10W off that, but 30W really is the limit if you want decent performance

You could also run VirtualBox on unRAID and install a Linux/Windows virtual machine for all your background/download applications.

This can be done very simply without changing your unRaid system. Simpler that going ESXi, in my opinion, if you don't need to.

Or just use it to experiment.

 

I vote for ESXi too. You've already got the hardware for it. Just follow the instructions in the Atlas thread and you'll be good to go. Set up a VM with Ubuntu and run jdownloader and other apps in that. Thats what I'm doing now.

My server and pc both have a base footprint of 80 watts, I cannot get it any lower.  am now using S3 sleep for the server so it remains on with disks spun down.

My server and pc both have a base footprint of 80 watts, I cannot get it any lower.  am now using S3 sleep for the server so it remains on with disks spun down.

 

With the similarities to my configuration, I wonder why you are 20Watts up on mine ... second SATA controller?  Additional drives, even when spun down?  Fans?

 

What PSU are you using?

 

While I did experiment with putting the server to sleep I abandoned the idea for two reasons:

 

1) With SqueezeboxServer, Mail server and constant downloads, it wouldm't sleep much anyway.

2) I'm concerned about clean shutdown when the power fails - would the machine wake up just to shut down?

 

As an example, we've just had three 20 minute power cuts in the last two hours, so this really is a big concern.

 

My UPS battery has been exhausted by these cuts, so I have left the server off for a while to allow the battery to recharge.

 

 

I've had a quick read-up on ESXi - while it sounds like a fun exercise, I have concerns about disrupting my server - again, would the UPS work okay, etc.

Also, it seems that it needs a dedicated (and fast?) Datastore device, and most people appear to be using SSDs in new builds ... or could I partition my cache drive for this purpose?

 

Perhaps I'll save up for an SSD to give it a try - I understand that new devices have a much better life expectancy, so I could always use it as my cache drive if I stick with a bare metal environment.

That 80 watts is for my complete setup, so server, pc, two monitors, printer, switch, and all as down on power as I want to go.

 

I am kind of an energy saver and I have been able to shave off hundreds of my electricity bill with several saving actions. I however have one base rule; I do not want to go down in usability and comfort. I would be able to save more but that would have impact in comfort.

I've just spent a few minutes measuring the power consumption of some of my IT equipment.  I don't have anything fancy like a Kill a watt (guess I should see whether I can get one which works on 240V/60Hz), but I'm just using the figures from my APC UPS (reads to the nearest 4W).  I'm sure that others have reported on similar experiments, but I wanted to add my findings.

 

 

You might be able to sign out a kill-a-watt meter from the public library if you are in the u.s. A few years ago there was some sort of energy saving program being promoted and many states were given the meters.

I've just spent a few minutes measuring the power consumption of some of my IT equipment.  I don't have anything fancy like a Kill a watt (guess I should see whether I can get one which works on 240V/60Hz), but I'm just using the figures from my APC UPS (reads to the nearest 4W).  I'm sure that others have reported on similar experiments, but I wanted to add my findings.

 

 

You might be able to sign out a kill-a-watt meter from the public library if you are in the u.s. A few years ago there was some sort of energy saving program being promoted and many states were given the meters.

 

 

If memory serves he is in the Philippines. As a legacy of U.S. colonial rule they use our electric sockets though not at the same voltage/frequency.

If memory serves he is in the Philippines. As a legacy of U.S. colonial rule they use our electric sockets though not at the same voltage/frequency.

 

Correct - for some strange reason they've adopted 230/240V here, but stuck with 60Hz.  Unfortunately, frequency conversion is more difficult than voltage conversion.  This means that most of my UK equipment will work, except those items with a synchronous motor - my problems are with tape deck, record deck and washing machine.  The washing machine is a strange one - the main motor is electronically controlled, all the solenoid valves etc are fine ... it's just the water pump which won't start at 60Hz.  I've found that removing about 20% of the winding gets it to start most of the time, but then it gets hot and thermal protection cuts in.

 

But, yes, back to the original point, I need a Kill a Watt which works on 240V.

How about this?

http://www.ebay.com/itm/NEW-USA-energy-meter-Watt-Voltage-Volt-Meter-Monitor-Analyzer-with-power-factor-/251163129167?pt=LH_DefaultDomain_0&hash=item3a7a7d354f

 

How about this?....

 

Ummm .... "Operating voltage 120V", and Paypal doesn't work for me.

There are also 220-240V versions.

 

I'm not aware of a 220v version of the Kill-a-Watt (the manufacturer's site doesn't show any such alternative).

 

HOWEVER, this very similar meter should do the trick:  http://www.ebay.com/itm/Advanced-WATT-Power-Energy-Meter-Monitor-220V-AU-PLUG-/261025164435

 

By the way, if you want a very low power UnRAID server, build one with the following.  My 2nd UnRAID server uses these components (in a slick little Lian Li PC-Q25B case) with 5 3TB WD Red drives.  Consumes about 20w when idle (drives spun down);  and just over 40w with all drives spun up and busy (during a parity check).    And very good performance -- ~ 115Mb/s data transfers; 35Mb/s writes;  and just over 8 hours for a parity check.

 

SuperMicro X7SPA-H-D525-O motherboard (with Atom D525)

4GB of RAM

5 3TB WD Red drives (the Lian-Li case has 5 hot-swap bays ... I mounted the parity drive on the bottom tray; so have room for one more 3TB drive if I ever need it)

 

By the way, if you want a very low power UnRAID server, build one with the following.  My 2nd UnRAID server uses these components (in a slick little Lian Li PC-Q25B case) with 5 3TB WD Red drives.  Consumes about 20w when idle (drives spun down);  and just over 40w with all drives spun up and busy (during a parity check).    And very good performance -- ~ 115Mb/s data transfers; 35Mb/s writes;  and just over 8 hours for a parity check.

 

SuperMicro X7SPA-H-D525-O motherboard (with Atom D525)

4GB of RAM

5 3TB WD Red drives (the Lian-Li case has 5 hot-swap bays ... I mounted the parity drive on the bottom tray; so have room for one more 3TB drive if I ever need it)

 

 

Supermicro actually has a successor line to those boards coming out. They use Intel's new Atom S1200 line of CPUs that were intended for server use from the beginning. And they're more power efficient than the D5xx line of Atoms. Unfortunately the first board released in the series only has 4 on board SATA ports and a normal 32-bit PCI slot so you're limited to 6 drives when you put in a 2 port controller card. Hopefully they release more soon because these will make great choices for low powered unraid systems. The chip on the board has 8 PCI Express 2.0 lanes so this thing paired with an IBM M1015 controller card could make a super efficient 12 drive array.

 

http://www.supermicro.com/products/motherboard/ATOM/X9/X9SBAA-F.cfm

My system consumes a little more power because of the Core i3 as opposed to an Atom, but still very low power.  33W at idle with all drives spun down, 69W with 5x 7200 RPM drives spun up and busy (parity check), and also has excellent performance.  Greens or Reds would obviously lower the peak consumption number as they require less power.  Build thread (link in sig) has full power consumption and performance numbers in OP.

Nice to know about the newer Supermicro option -- but quite frankly I'd stay with the D525 board if I was building another unit right now, since it has 6 SATA ports .. which is perfect for the little PC-Q25B case.

 

However, I'm also intrigued with the excellent power consumption of the Asus H77 board with an i3 => if you want to do anything besides just basic UnRAID that's a very attractive option, as it has a LOT more CPU "horsepower" than an Atom.    If you just want an UnRAID storage server, however, the D525 has PLENTY of "horsepower".

Nice to know about the newer Supermicro option -- but quite frankly I'd stay with the D525 board if I was building another unit right now, since it has 6 SATA ports .. which is perfect for the little PC-Q25B case.

 

However, I'm also intrigued with the excellent power consumption of the Asus H77 board with an i3 => if you want to do anything besides just basic UnRAID that's a very attractive option, as it has a LOT more CPU "horsepower" than an Atom.    If you just want an UnRAID storage server, however, the D525 has PLENTY of "horsepower".

 

 

There is actually a low powered Xeon processor that is rated at only 17 watts. The D525 is rated at 13 watts. If money wasn't a big issue, since this would require server grade motherboard and RAM which cost a bit more, that is the way I'd go for sure. In comparison the low powered i3 Ivy Bridges are 35 watts and the normal i3 Ivy Bridges are 55 watts.

 

http://ark.intel.com/m/products/65735/

 

Intel makes a Mini-ITX board compatible with the chip.

 

http://m.newegg.com/Product/index?itemNumber=N82E16813121743

There is actually a low powered Xeon processor that is rated at only 17 watts. The D525 is rated at 13 watts. If money wasn't a big issue, since this would require server grade motherboard and RAM which cost a bit more, that is the way I'd go for sure. In comparison the low powered i3 Ivy Bridges are 35 watts and the normal i3 Ivy Bridges are 55 watts.

 

Comparing maximum power figures when most users are concerned about power consumption when idle may not give the best indication.  i3 Ivybridge CPUs idle at quite low power.  The system power requirements can then become significantly influenced by the motherboard chipset, RAM, any add on cards, the number and types of drives, the number of fans (and whether they are running at reduces power), and the efficiency of the power supply at very light loadings (which is often significantly worse that the quoted efficiency).  My "normal" i3system (8 drives, spec in my sig.) idles at around 38 - 39 watts.

Agree -- the chipset, fans, and especially PSU efficiency at very low loads can make MUCH more difference than the max power rating of the CPU.    For example, the very low power Xeons and i3's use almost exactly the same power at very low CPU utilization as their more power-hungry big brothers ... they are simply throttled to ensure they never use more than the max rating.

 

Power supplies running at less than 20% load can have VERY steep drop-offs in their efficiency ... so even an 80+ Platinum unit won't be very efficient  [Remember, 80+ certification is done at 20%, 50%, and 100% load points].    So even a 300w PSU will be operating well below its rated efficiency at < 60w consumption.    I suspect the reason my system's load only goes up 20w when I switch from idle (~ 20w) to max power with 5 drives spinning and doing a parity check (~ 40w) --- which according to the drive's specs should cause about a 25w increase -- is because the PSU is slightly more efficient with the higher load.

 

I think the idle load is much more influenced by the chipset, motherboard components, fan, and PSU efficiency than the specific CPU, since almost all modern CPUs draw very low power when throttled down.

 

... and of course nobody makes an 80+ certified 100-150w PSU, which would be ideal for these very low power systems => especially in an SFX form factor 

There is actually a low powered Xeon processor that is rated at only 17 watts. The D525 is rated at 13 watts. If money wasn't a big issue, since this would require server grade motherboard and RAM which cost a bit more, that is the way I'd go for sure. In comparison the low powered i3 Ivy Bridges are 35 watts and the normal i3 Ivy Bridges are 55 watts.

 

Comparing maximum power figures when most users are concerned about power consumption when idle may not give the best indication.  i3 Ivybridge CPUs idle at quite low power.  The system power requirements can then become significantly influenced by the motherboard chipset, RAM, any add on cards, the number and types of drives, the number of fans (and whether they are running at reduces power), and the efficiency of the power supply at very light loadings (which is often significantly worse that the quoted efficiency).  My "normal" i3system (8 drives, spec in my sig.) idles at around 38 - 39 watts.

 

Well if you can find me statistics that ECC RAM and a C206 chipset use that much more power than a desktop chipset I'll buy in to your point. I think it's a fair assumption to make that if Intel says it will use less power under load it will use less power at idle as well. This is the same thing they do with their mobile chips. Intel mobile chips are just desktop chips binned as mobile. Intel doesn't bin them as mobile just because they use less power at load, but because their whole power consumption profile is lower than a desktop part. I think it's fair to assume the same with their "L" version desktop chips as well.

My point was that a chip that has, for example, a tdp figure of 35 watts may have an idle figure of 5 watts or so.  At that level the other parts of the system also make a big difference.  Yes, a processor with 17 watts tdp is likely to idle at lower power than a chip with 35 watts tdp if they use the same silicon manufacturing node and processes, but unless you take care over every other part of the system, the real benefits may not be achieved.

 

On your other point, I cannot speak for Intel (I was a Texas Instruments employee), but I doubt that "Intel mobile chips are just desktop chips binned as mobile.".  Substantially lower power requires design and manufacturing process changes as well as "binning" during the production  flow.