Which setting is the better OC?

[TJCS]

Rig is my sig. ASRock Z77 Extreme4 BIOS 2.00.


What do you think? Setting A or B

 

[TJCS]

I guess I will take a stab...

According to Anandtech's 2007 LLC review, Setting A is the better option by allowing CPU to control the droop and peak voltage. But shouldn't Setting B's voltage be higher since LLC is set at max? Is CPUZ not reading the voltage correctly that's why temps are so close?

 

[TJCS]

Did I make the chart too complicated? Missing some info? TELLLLL MEEEE or HEEEEEELLLLPPPPPP MEEEE

 

[Idontcare]

I guess I will take a stab...

According to Anandtech's 2007 LLC review, Setting A is the better option by allowing CPU to control the droop and peak voltage. But shouldn't Setting B's voltage be higher since LLC is set at max? Is CPUZ not reading the voltage correctly that's why temps are so close?

Your chart is fine, but that review is beyond outdated, way beyond outdated.

Even at the time of its publication, it was essentially the pinnacle of "in search of the academically correct answer, sans all that pesky practical stuff that happen in reality".

Technically correct but lacking in all the ways that matter when it came down to the nut of it.

The differences in LLC settings, once the offset has been optimized as you have done, is essentially nada. It is six of one versus a half-dozen of the other.

Technically the processor's lifetime will be different depending on which option you choose, but you are really splitting hairs at that point and are dickering around in choosing between the option that yields 9.57632yrs of lifespan versus the option that yields 9.57631yrs of lifespan. (figuratively speaking of course)

The significant digits of interest here are the ones that precede the decimal place, your processor is going to last years and years at those voltages and operating temperatures, neither option is going to rob you of your CPU at an inopportune time

As for CPUz, yes it quantizes the reported voltage in 0.008V increments, and rounds down (under-reports) to boot

FWIW the formula for computing the CPUz value based on knowing the actual value is as follows:

In Excel the formula is as follows:

=ROUNDDOWN(([COLOR=darkgreen]A1[/COLOR]/0.001)/8,0)*8/1000

^ Where "A1" is the cell containing the actual Vcc per the BIOS (or multimeter if you have one setup).

 

[TJCS]

Thank you IDC, I was hoping that you show up to clear this up ^_^ I think sometimes when I read too much info on the net, it serves more confusion than good.

This has been boggling my mind for a few days...

I also wanted to ask:

1. My two LLC settings are set at two extreme ends, and the one that idles around VID voltage is the one where LLC is off, is that correct?
2. I shouldn't be worried about VRMs overheating or exploding if I run the above setting with max LLC 24/7 right? (I ask because I blew up my VRMs on my Asrock X58 sometime ago, but the voltages involved in that incident was much higher).

 

[Idontcare]

No problem

I really can't say with any confidence which LLC setting is the closest to being "no LLC" because of the offsets involved.

Setting A is closest to the VID at idle but that is only because of the +0.04V offset you have there.

If you consider that the Vcc for Setting B is being undervolted by -0.06V then we would expect Setting B w/zero offset would yield an idle voltage that is nearly identical to the VID value.

Wish I could answer this better for you, I have no excuse either because I've got an Asrock z11 extreme 6 mobo sitting in the box waiting for me to dig it out and play with it but I haven't gotten to it yet. So if you do find out which way the LLC settings go then I'd very much like to hear back from you on that. Educate me, please!

 

[TJCS]

No problem
If you consider that the Vcc for Setting B is being undervolted by -0.06V then we would expect Setting B w/zero offset would yield an idle voltage that is nearly identical to the VID value.

It's insane that you pointed that out... I completely dismissed that setting because I was doing 1.152-0.06... oh man should've paid more attention in school...

Hmm, don't know of a z11...hahah, and there happens to be a Extreme11 and a Extreme6. This review confirms LLC is strongest on level-1 for Extreme6.

For Extreme4, however, this AT thread claims it's backwards.

I also confirmed earlier this week with Yuriman:

Prior BIOS 2.30: LLC Level-5=LLC 100%
After BIOS 2.30: LLC Level-5=LLC OFF

Edit:
I think I found which level is strongest on my Extreme4:

+0.04 L1 Idle(1.072, 1.080) Load(1.248, 1.256)
+0.04 L5 Idle(1.064, 1.072) Load(1.152, 1.160)

 

[Vectronic]

The Z77 Ext4 LLC was "fixed" with a BIOS after 2.30 (can't say which, probably 2.50, changelog doesn't say)... I've used 2.5 and 2.7 quite a bit, both behaved "normally" on two different Ext4s.

I say "fixed" and "normally" because LLC on the Ext4 is useless really... it's almost as bad as "pick a number, see what happens".

That said, and what has already been said, just use whichever one grants you the most stability with the least temperature.

I highly suggest you don't investigate the Ext4 too extensively, the more you look around, the less confidence you'll have... just play with it.

 

[TJCS]

Thanks for the inputs guys. I have decided to use LLC Level 5

Since there is not much temp difference between the two setting, I think LLC is unnecessary in this situation.

Now I know why VRMs are going to be integrated into haswell

In the future, we won't need to worry about exploding VRMs... just exploding CPUs...