Xeon 26xx v2 power draw

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

darkneika

New Member
Nov 16, 2016
2
0
1
30
hi guys,
I'm about to build a home server based on unraid or freenas, I would like to know the consumption for the following build in idle:
e5 2660 v2
8x8 ddr3
lsi 9201 16e
1 tb samsung evo
6x8 wd reds
the consumption could be 100w in idle ??
Thanks in advice!
 

acquacow

Well-Known Member
Feb 15, 2017
784
439
63
42
If you enable c-states and other power saving features in the bios, you might be able to get it down to 100W. I have an E5-2648L (70W TDP) system that idles just at 100W with a little bit more hardware in it.

Your 2660 is 95w TDP, but it should be able to idle down to roughly the same power states.
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
That's alot of server for freenas :D and will use >100w at idle with all that.
 

am4593

Active Member
Feb 20, 2017
150
35
28
44
I read somewhere that some people see major power savings running unraid instead of freenas. Has anyone run across a direct comparison of power savings between the two OS. I run freenas with an E5 1650v2. Idle is somewhere around 150w, which hurts. 128GB ARC and connectx3. Since that machine mostly stores media files I would consider moving to unraid if the power savings were good.
 

Skud

Active Member
Jan 3, 2012
150
78
28
Our DL360 G8s with dual 2690 v2s and 64GB DDR3 idle around 75-90w in Windows according to the iLO.

They have 4x600gb 10k SAS and dual 750w PSUs.

Not that it is relevant but today I just decommissioned a dual X7xxx something DL580 G7 with 128GB RAM and 4x600 GB that was idling at 475w.

EDIT: I should also mention that we run everything on 240v.

Riley
 
Last edited:

am4593

Active Member
Feb 20, 2017
150
35
28
44
my understanding is that because unraid only access 1 or 2 disk at a time instead of a whole array there are real potential power savings if you have a large array
 

kapone

Well-Known Member
May 23, 2015
1,095
642
113
Lots of misconceptions in this thread...

@Skud is exactly right. It depends on the motherboard/BIOS settings/fans/chassis/power supplies etc etc, not including RAID cards and drives. That has a major impact.

With 2x Ivy Bridge CPUs and ~64GB RAM and a few fans, with the motherboard power settings set to "the equivalent of low power" (depending on the motherboard) you're gonna see ~75-80w on average. That may go up or down a bit depending on individual circumstances, but the deviation won't be a major factor. This assumes generally good hardware, like PWM fans, platinum rated PSUs, motherboards with features etc etc.

RAID cards have a definable and measurable power draw. That's not gonna change regardless of the OS. Just add that to the above figure.

Now...drives and OS... IF and that's a major IF, you don't spin down your drives, the power draw will be similar across almost all major OSes. If you do spin down, and assuming the OS allows you to, that's where you'll see the difference.

re: freenas vs unraid - You're talking apples and oranges. Actually more like Madonna and Miley Cyrus if you catch my drift. :) Two very different things. One is ZFS, the other one is not. Pick your poison.
 

am4593

Active Member
Feb 20, 2017
150
35
28
44
well aware of the differences in freenas vs unraid but that doesnt really answer my question. They both have different use cases. You'd be crazy to store stuff that you really cared about with unraid, but for large media library it could be useful for large power savings, i'm just curious what exactly the power savings look like since unraid will spin up fewer drives to access data.
 

kapone

Well-Known Member
May 23, 2015
1,095
642
113
well aware of the differences in freenas vs unraid but that doesnt really answer my question. They both have different use cases. You'd be crazy to store stuff that you really cared about with unraid, but for large media library it could be useful for large power savings, i'm just curious what exactly the power savings look like since unraid will spin up fewer drives to access data.
Unraid will spin up exactly ONE drive (assuming you're spinning down drives, and we're talking media server type usage) depending on the current media playing.

So...I'm still not sure what the question is. Apples and Oranges (or Madonna and Miley Cyrus) still apply.
 

am4593

Active Member
Feb 20, 2017
150
35
28
44
quest has to do with what kind of power savings are implied between unraid and freenas. surely someone here has done a comparative analysis of their experience using the two and the power draw that they've got. love my freenas box but 150w idle hurts.
 

kapone

Well-Known Member
May 23, 2015
1,095
642
113
quest has to do with what kind of power savings are implied between unraid and freenas. surely someone here has done a comparative analysis of their experience using the two and the power draw that they've got. love my freenas box but 150w idle hurts.
I don't wanna sound like a broken record...but... :)

150w idle is with what? all drives spun down?

If that's the case, with the exact same hardware, the idle power usage is not gonna change substantially.
 

am4593

Active Member
Feb 20, 2017
150
35
28
44
so I assume you request any data from a ZFS array all of the disks will spin up. in my case thats 10 spinners. Thats x number of minutes of use where all 10 are running + x number of idle minutes before they all spin down. Unraid, spin up 1 + maybe paridy disks, same use case. I would think there would be power savings in that
 

EffrafaxOfWug

Radioactive Member
Feb 12, 2015
1,394
511
113
Depends on how much your power costs I guess, but 6x spinners isn't a great deal of platters to keep spun up all the time - my 24x7 server uses only six spinners and they eat about 5W each idling, so ~30W overall. To me this is acceptable because from my own experience I suffer significantly higher failure rates on drives that are spun up and down all the time than ones that are constantly running so any saving in power is negated by extra expenditure on HDDs. If I needed to use more discs I would probably re-think my strategy (or do some more testing on spin-up). But I do find the 5-10s delay waiting for a drive/all drives to spin up maddeningly frustrating and it lowers the WAF as well when something doesn't load instantly. My £0.02, YMMV.

As for the OP, as everyone else has pointed out it's "how long is a piece of string?" but I don't think there's any way that you'll get two IVB Xeons and 8 DIMMs in under 100W and 150-200W idle is more likely. IIRC the v2 Xeon's had significantly higher idle draw than the v3 (Haswell) Xeons; I seem to remember a single v2 drawing around 15-20W more at idle than a comparable v3, more so on 2P boxes.

Fans are also capable of drawing a whole heap of power - 10-20W each isn't unusual for yer hi-falutin' enterprise screamers.

The LSI 9201-16e - IIRC this is one of the "two chips glued together" generation so it will have significantly higher power draw than a 9201-8e - as per the ye olde STH HBA power comparison it'll draw 15-20W, twice that of the 8-port HBAs. I'm curious as to an external HBA though, are the discs housed in an external JBOD or are you using an internal loopback?

If you're looking to keep power to a minimum, I'd start off seeing if one CPU would be viable (and unless you're doing something exotic with FreeNAS I think it should be - I've got a backup server running ZFS on a Xeon E3-1230 v2 and it's never short on CPU power), and you'd be better off going for a Haswell-or-newer to cut down on idle power, along with a case you can cool with some low-powered fans if possible.
 
Last edited:

DRAGONKZ

Member
Apr 9, 2018
87
10
8
41
I’ve got 3 servers, all running esxi 6.7.

1. HP DL360P G8, 2 x E5-2640 v2, 16x16GB 1866 DDR3, P420i, 1 x SAS SSD, 2 x SATA SSD = 96 watts

2. Cisco C220 M3, 2 x E5-2697 v2, 16x16GB 1866 DDR3, 9271 RAID, 10Gb VIC, 4 x SAS SSD, 4 x SAS 10k = 236 watts

3. Cisco C220 M3, 2 x E5-2697 v2, 16x8GB 1600 DDR3, 9271 RAID, 10Gb VIC, 2 x SAS SSD, 4 x SAS 10k = 142 watts

Server number 1 runs 2 VMs, server 2 runs 22 VMs, server 3 runs no VMs. (The environment is still being built)

Biggest increase in power wattage in my experience comes from CPU, RAM, spinning disks, and actual real world usage.

FWIW, I had HP DL20 Gen9s before these and with them completely maxed with every option to the highest spec they ran VM workload at around 25 watts usage per server!

They were very impressive power usage wise.
 
  • Like
Reactions: Evan

kapone

Well-Known Member
May 23, 2015
1,095
642
113
so I assume you request any data from a ZFS array all of the disks will spin up. in my case thats 10 spinners. Thats x number of minutes of use where all 10 are running + x number of idle minutes before they all spin down. Unraid, spin up 1 + maybe paridy disks, same use case. I would think there would be power savings in that
But that's not idle power consumption. That's power consumption, while using.
 

neb50

Member
Aug 28, 2018
73
23
18
Dell R720xd with 2 - e5-2650v2's, 8x8 PC3L, Intel 2-10G card idles as low as 80w without raid/hba card or drives.
Add in a 9201-8i, 4 - 8tb WD red drives, 2 - 1.6tb intel S3500 SSD's, and a Intel optane 16Gb drive with adaptor bumps it to around 125w
Add in 3-4 vm's with one running BlueIris with 9 cameras will bump it up to around 160-175w

This is with ESXi6.5 with FreeNAS running as a VM along with a couple Windows and Linux VM's

The spinning drives take around 4-5w each to keep them spinning, the HBA is around 10w, and each of the 10G links adds a few watts

Keeping the drives spinning is around $20 per year for me and it is worth it to not have to wait for the drives to spin up or possibly fail early.
 

Skud

Active Member
Jan 3, 2012
150
78
28
Some more numbers and comparisons...

DL360 G7 - 2 x E5640 - 64GB - 2 x 300GB SAS - Server 2008R2 - Idle: 110w

Same server as above but with load - 75User Citrix Shared Desktop - (75%CPU, 80% RAM): 230w

DL360p G8 - 2 x E5-2640V2 - 64GB - 2 x 600GB SAS - Server 2008R2 - 75User Citrix (75%CPU, 80% RAM): 190w

Also, I forgot to mention - we run all our servers @ 240v. So, expect 120v numbers to be a few % higher.
 

sovking

Member
Jun 2, 2011
84
9
8
There is a lot of confusion in this thread.
Regarding the OP's question, I think you could stay around 100w.

My home server is composed by: SC836, X9srl-F, E5-2680v2, 64 GiB DDR3 RDIMM, LSI 3008-8i, SAS3 expander backplane, 4x4TB WD RED, 1 ssd, no fans. It load ESXi 6.7, with Freenas and other 3 VM. It idles around 90W.
Accessing freenas from common operations does not increase a lot power consumption: it remain around 100-110 W.
Spinning low rpm disk use around 5w each, that is quite good.

In my opinion, if your primary usage is as NAS server of only 4-6-8 disks, you could reconsider your hardware list: use a motherboard for xeon 4-core haswell or successor, with no hba, no expander. You will get not more than 50 W idle, and more performance for sharing due higher clock.
If your primary usage is virtualization, getting certifications, rendering, nas with many drives (more than 8), use Xeon E5 ivy bridge or successor, single or dual cpu, depending or your real needs and budget.

You can even get 1st solution for low-power 24/7 nas and add 2nd solution to power-up only when you need to do your work/experiments.

Generally, the power consumption depends on hw you are using; not much on O.S. (at least in x86 world)
 

darkneika

New Member
Nov 16, 2016
2
0
1
30
so it would be better a system based on ryzen 2600 or a xeon 1245v6, the best cpu would be the new xeon e2176g but the price is really prohibitive, with the two systems mentioned above I would be able to have a good energy savings?
the main use of the server will be plex media server , dockers and one or at most 2 vm and pfsense
 
Last edited: