E5-2670 - has anyone disabled cores?

[MiniKnight]

Spawned by the dual socket FreeNAS thread, has anyone disabled cores via BIOS on the E5-2670 and seen if there was an actual power savings? I'd suspect little if any except on the basis of a maximum power draw. The E5-2660 and E5-2670's are so cheap I'd consider running 2 sockets with 4 cores active just to get more RAM and PCIe slots. Using lower density RAM could make up the $55-60 on the second chip.

 

[Zack Hehmann]

@_alex posted this in this thread Is it worth to replace E5 2670 with 2630L ?

Hi,
this is what i observed by disabling cores in the BIOS - absolutely nothing changed in idle-consumption.

The two things that made me think it could be worth to go for 2630L on this node are:
- Lower count of Cores, 6 vs. 8 - could save 25% in theory

- here, at 'typical power consumption', there is a difference of nearly 50 watts (52% of the 2670).
So, i wonder if this is true and where these numbers come from:

Intel Xeon E5-2670 vs E5-2630L

Alex

 

[Joseph Nunn]

What I might try in your position is disabling hyperthreading if 16 threads is all you need rather than disable physical cores. You'll reduce the maximum threads, power draw, and heat, but without leaving your real computational power on the table. My CPUs are not here yet so I can't test, but I believe idle power draw will likely not be affected by disabling cores.

Joseph

 

[Patrick]

@MiniKnight - I think what @Zack Hehmann and @_alex mentioned is right. I have actually done this on some Xeon D Ceph nodes but it did not reduce idle to a significant degree. It did, however, lower temps significantly at load.

@Joseph Nunn I suspect you are right on the idle power draw. HT is generally a high performance/ watt capability.

 

[Joseph Nunn]

@MiniKnight - I think what @Zack Hehmann and @_alex mentioned is right. I have actually done this on some Xeon D Ceph nodes but it did not reduce idle to a significant degree. It did, however, lower temps significantly at load.

@Joseph Nunn I suspect you are right on the idle power draw. HT is generally a high performance/ watt capability.

Found this chart on Ivy Bridge:

http://blog.stuffedcow.net/wp-content/uploads/2012/07/ivb_power.png

I can understand HT might not be the biggest power draw, which is likely to be dominated by the functional units themselves rather than the second copy of the registers HT uses. However HT cannot be used effectively for all workloads, as in the case of concurrent vs truly parallel jobs, whereas real cores can. Therefore I would be inclined to disable HT before considering disabling cores in an effort to reduce power draw. However the chart I found above does indicate that disabling cores is a better approach for truly reducing power draw than disabling HT.

Joseph