Just curious what people are getting for energy usage on their servers. I just got one of these Kill-A-Watt deals and have been checking stuff. My Unraid box is using 121.8 watts all drives spun down. Funny thing is that when I spin up all drives (4 SATA and 3 IDE) it says it is using exactly the same. My normal windows machine is in the 115 area. Wondering if I would see any power savings with the RB-1200 server

 

This all translates into $8.40 a month to run it.

My Unraid box is using 121.8 watts all drives spun down.

when I spin up all drives (4 SATA and 3 IDE) it says it is using exactly the same.

 

As this can't possibly be true, it means something's wrong with your Kill-A-Watt.

Try measuring different light bulbs. You know their wattage.

 

 

After it sat for awhile now about 30 minutes I see that it is down to 96 watts and jumped to 144 when all spinning. Spun them down and dropped right back to 96.

 

I don't know, I didn't tuch it. Maybe it was just stuck for a bit 

I have an 8 SATA drive server, router, 5-port GBswitch, and DSL modem all connected to an APC UPS and the UPS output reports around 100 watts.

My system draws 53 - 56 watts during a parity check, around 70 watts during parity check and 100% CPU usage during a parallel kernel compile, and around 30 watts when idle. If you want more details, search the other threads on this subject.

dropped right back to 96.

 

In my experience, that's very high. I have a handful or so of computers that are or are similar to an unRAID server, and none use more than 60W at idle.

 

I've only ever seen >100W @ idle on computers with dual graphics cards.

 

For example, I have an OpenSolaris workstation that idles around the numbers you quoted first: 120--130W. It has a quad-core CPU, 8GB RAM, 6 (old!) IDE HDDs, 3 SSDs, 1 SATA DVD drive, an 8600GT GPU, and a 9600GT GPU.

 

 

Are you by any chance running the server with an older, inefficient graphics card and all on-board hardware enabled in the BIOS?

My Abit AB9 Pro with 9 drives idles is around 90-110 watts at low usage (don't remember the exact amount).

This is with 2 7200 RPM 32MB cache 1.5tb Seagates running 24x7 with rtorrent.

The other drives are 5400 green drives which are spun down until I access movies.

It is an old Asus P4T533-C mother board (no on board SATA). The NIC is onboard and that is it. I don't even remeber what the graphics card is but it was a gammer card at the time. Older Antec case with 3 fans plus a fan on the 3 in 1 cage. Pentium 4 2.26gz.

 

The fans run full speed all the time, no control for them.

 

I knew that this wasn't a very efficent set-up, I was just wonder how bad it was.

My AMD 740G/BE2300 combo runs something like 55W idle and maybe 70W with the drives running. It's been a while and I have changed power supplies from an old piece of junk to a Corsair since I measured it so I should measure it again.

 

To compare, my Nvidia 9300/Intel E6300 mini-ITX system runs about 38W idle with the hard drive is still spinning using a brick type power supply. This has proved to be a great HTPC so far but heat can sure be a trick to manage in a mini-ITX package. The new power supply dropped the measured wall power by about 10W compared to the supply that came with the case.

 

Peter

Can you list how many drives you have? I have a server with 18 drives. That uses over 100W on idle.

I have 4 drives in my server right now - the AMD system. The mini-ITX is the HTPC with one hard drive.

 

Any newer large capacity SATA drive should use no more than about 1W when spun down. So, 18 drives should use no more than 18W when they are all spun down.

 

Peter

 

I have three SATA drives in my tower, and a Gigabyte GA-MA790GP-UD4H board with an AMD Athlon 64 X2 5050e Brisbane 2.6GHz. With two fans for the HDs and a CPU fan, I measure 41W when the drives are spun down and 58 W with all drives up and running. It used to be running around 70W, but I underclocked / undervoltaged the system (5x multiplier with 200MHz FSB and 0.8 V for the core) the other day to reduce the power consumption, which seemed to have worked.