Load Line Calibration

[jjrowley]

No matter what o/c I go with, I cannot get it stable with LLC disabled. I can get 4003MHz stable with it on, but voltage is like 1.4250. I can get 3.8 stable at 1.38 with it on. I'm at 3.6 and stable with it on at 1.3500. If I turn it to disabled none of them are stable. what gives?

 

[PlasmaBomb]

Hmmm... that is odd.

As we have discussed before Loadline Calibration is best left disabled, especially when using the newer 45nm CPUs. Our testing has shown these settings induce power instabilities, even when using lower voltages. In addition, when enabled the option sometimes requires more CPU voltage than would otherwise be necessary, meaning increased power dissipation and higher temperatures.

http://anandtech.com/mb/showdoc.aspx?i=3208&p=2
http://www.anandtech.com/cpuch...howdoc.aspx?i=3184&p=6

A quick google -

This seems to be the answer - LLC boosts the Vcore

http://www.hardwarecanucks.com...erboard-review-14.html

(bottom graph)

 

[Concillian]

Are you increasing voltage so that you get the same reported voltage in CPUz as with it enabled?

If you enable it and leave the same voltage in BIOS, your reported output voltage will decrease. You need to increase the voltage in BIOS until you get the right voltage reported by CPUz or HWmonitor or whatever you are checking it with.

 

[Idontcare]

I'm not entirely sure why Anandtech tends to be officially down on LLC so much in their reviews but personally LLC makes all the difference in my ability to hit a 24x7 stable 4GHz overclock on my QX6700 under phase.

If I take it off LLC then I have to push nearly an extra 0.1V (1.60V versus 1.50625V) thru the chip to keep it small FFT stable at 4GHz...which is then not viable for 24x7 operation at full load because the extra voltage at that GHz pushes the total power consumption above my vapochill's refrigeration capacity.

I agree LLC is not the only way to successfully and safely overclock, but I haven't quite figured out why LLC is vilified as much it seems to be across the web.

I've encountered no reportings of LLC actually killing a chip, and you can bet it would have long ago disappeared from ASUS mobo BIOS options if it were in fact fingered by Intel's returns dept as a highly likely suspect were any rash of in-field chipkills occurring from users with Asus mobos in their rigs.

So...I guess it just feels good sometimes to be haten on something, and LLC generously stepped up to fill that role quite nicely.

 

[dust]

I had the same problem while having the LLC disabled, I had to raise the voltage quite a lot for an extra 100 mhz and it's not worthed. I saw many complaining abt the LLC, once I have enabled it though, I could apply lower increments in the voltage to make the system stable. For instance, for the rig in my sig, with LLC disabled I need 1,4v vcore in Bios to get as low as 1,37 reading in CPU-z, whilst with it enabled I only have to put 1,38v and the system is stable.

I would need to consider raising other voltages as well with the LLC disabled, like the FSB V termination and the CPU PLL V. It seems that I need it enabled to get a better margin and I'm ok with it, I'm running a multi limited 45nm quad with high VID(1,25), therefore I have little room for improvement. I would really like to know myself what is it so wrong abt LLC, I couldn't find anything so far.

 

[dust]

Originally posted by: Idontcare
I'm not entirely sure why Anandtech tends to be officially down on LLC so much in their reviews but personally LLC makes all the difference in my ability to hit a 24x7 stable 4GHz overclock on my QX6700 under phase.

If I take it off LLC then I have to push nearly an extra 0.1V (1.60V versus 1.50625V) thru the chip to keep it small FFT stable at 4GHz...which is then not viable for 24x7 operation at full load because the extra voltage at that GHz pushes the total power consumption above my vapochill's refrigeration capacity.

I agree LLC is not the only way to successfully and safely overclock, but I haven't quite figured out why LLC is vilified as much it seems to be across the web.

I've encountered no reportings of LLC actually killing a chip, and you can bet it would have long ago disappeared from ASUS mobo BIOS options if it were in fact fingered by Intel's returns dept as a highly likely suspect were any rash of in-field chipkills occurring from users with Asus mobos in their rigs.

So...I guess it just feels good sometimes to be haten on something, and LLC generously stepped up to fill that role quite nicely.

Isn't 1,5v already a little bit too much?

 

[PlasmaBomb]

Originally posted by: Drivenbyvoltage

Isn't 1,5v already a little bit too much?

For a 65nm QX6700 1.5V is the upper end of Intels VID Voltage Range, so should be OK.

IIRC the maximum recommended is 1.65V for the 65nm process and 1.45V for the 45nm from Intel papers...

 

[Idontcare]

Originally posted by: PlasmaBomb

Originally posted by: Drivenbyvoltage

Isn't 1,5v already a little bit too much?

For a 65nm QX6700 1.5V is the upper end of Intels VID Voltage Range, so should be OK.

IIRC the maximum recommended is 1.65V for the 65nm process and 1.45V for the 45nm from Intel papers...

And in my case it was also on phase cooling, so -30C temps which helps make it a shade more tolerant to elevated voltages.

 

[jjrowley]

1.45v for the 45nm? Are you sure about this? Where do you get this info? If this is true I'm going back to my 4GHz overclock I had stable at 1.425 vcore. However everyone and there brother seems to believe the max safe voltage on a 45nm is 1.3625. I did happen to get stable at 3.4GHz with LLC turned off. But seriously, 4GHz is an epeen manifestation that I want to run stable. LoL.

 

[JAG87]

Originally posted by: jjrowley
1.45v for the 45nm? Are you sure about this? Where do you get this info? If this is true I'm going back to my 4GHz overclock I had stable at 1.425 vcore. However everyone and there brother seems to believe the max safe voltage on a 45nm is 1.3625. I did happen to get stable at 3.4GHz with LLC turned off. But seriously, 4GHz is an epeen manifestation that I want to run stable. LoL.

http://download.intel.com/desi...or/datashts/318726.pdf

page 19

 

[jjrowley]

Wow.. I have a whole new ballpark to play in now. Thanks for that. When I come back crying that you made me fry my chip, reference this post. LMAO!!!

 

[dust]

I had the chip at this voltage before and the temps reached 80c in about 12 mins on prime95 large fft test. During this time one side of the case was open and the ac in the room was blowing straight on the case( getting 2-3 c lower like this). I had six fans running at 100% in the case to maintain the air flow but at this vcc it didn't quite matter. I tried once setting the voltages on auto for everything with 450 FSB and I noticed the vcc was at 1,38v in Bios, instead the NB was set at 1,6v which seems quite high and the temps were again scary. Provided the air flow in the case is all right, it seems that the higher vcore can be set only with water cooling.

 

[jjrowley]

I had it stable at 1.4250 and 471FSB for 4003MHz reaching a max temp of 71C on prime overnight. I lowered it down because people kept telling me the voltage was too high. My NB wasn't anywhere near 1.6, it was like 1.4 max. I have a fan blowing in front of my tower but nothing like you are describing.

 

[JAG87]

Originally posted by: jjrowley
I had it stable at 1.4250 and 471FSB for 4003MHz reaching a max temp of 71C on prime overnight. I lowered it down because people kept telling me the voltage was too high. My NB wasn't anywhere near 1.6, it was like 1.4 max. I have a fan blowing in front of my tower but nothing like you are describing.

It shouldn't take 1.4250V for an E0 stepping cpu to reach 4.0, just FYI. It shouldn't take 1.38 for 3.8 nor 1.35 for 3.6. Either you have a really bad chip or you are doing something terribly wrong. What is the VID of your chip?

 

[n7]

Originally posted by: JAG87

Originally posted by: jjrowley
I had it stable at 1.4250 and 471FSB for 4003MHz reaching a max temp of 71C on prime overnight. I lowered it down because people kept telling me the voltage was too high. My NB wasn't anywhere near 1.6, it was like 1.4 max. I have a fan blowing in front of my tower but nothing like you are describing.

It shouldn't take 1.4250V for an E0 stepping cpu to reach 4.0, just FYI. It shouldn't take 1.38 for 3.8 nor 1.35 for 3.6. Either you have a really bad chip or you are doing something terribly wrong. What is the VID of your chip?

Actually, based on my results with my E0 Q9550, needing 1.425v makes sense.

I could not get 4 GHz stable with 1.4v no matter what i did.

I'd initially thought it was vTT/vNB/GTLs, & countless other things, but after getting 471 stable at lower multipliers but not at 8.5x, i realized it was vcore, at least for mine.
It also became obvious when dropping one notch below 1.4v of vcore @ my 24/7 3.95 GHz = unstable.

 

[n7]

jjrowley, LLC prevents vdroop.

If you disable it, you'd need over 1.45v (likely about 1.46v) in the bios to achieve around 1.425v under load.
With LLC enabled, it prevents vdroop...it actually bumps up vcore a bit under load.

So if you set 1.425v in bios, it'll likely be 1.42ish idle, 1.425v load.

 

[jjrowley]

Originally posted by: JAG87

Originally posted by: jjrowley
I had it stable at 1.4250 and 471FSB for 4003MHz reaching a max temp of 71C on prime overnight. I lowered it down because people kept telling me the voltage was too high. My NB wasn't anywhere near 1.6, it was like 1.4 max. I have a fan blowing in front of my tower but nothing like you are describing.

It shouldn't take 1.4250V for an E0 stepping cpu to reach 4.0, just FYI. It shouldn't take 1.38 for 3.8 nor 1.35 for 3.6. Either you have a really bad chip or you are doing something terribly wrong. What is the VID of your chip?

I'm more than certain its probably user error. I've tried to ask questions that I have, and have been fiddling with numbers here and there to see what effects what and what kinda goes together. I'll be the first to admit I am an amateur. This is my first build, and my first time *trying* to OC something. "I'm new, I'm new, I don't know what to do!"

What does VID even mean anyways? Core temp says VID:1.2500v and I've never seen that number change.

 

[jjrowley]

Originally posted by: n7
jjrowley, LLC prevents vdroop.

If you disable it, you'd need over 1.45v (likely about 1.46v) in the bios to achieve around 1.425v under load.
With LLC enabled, it prevents vdroop...it actually bumps up vcore a bit under load.

So if you set 1.425v in bios, it'll likely be 1.42ish idle, 1.425v load.

Yeah when I was setting at 1.425 I had LLC enabled. Does all this playing around, and stressing, and changing some more, and stressing, over stress these chips at a certain point where I am significantly decreasing its life. I understand OCing will shorten the life, but it is likely there won't be a time where I am ever running this thing under 100% load.

 

[JAG87]

n7,

it is usually a combination of motherboard and processor, but an E0 stepping cpu should not take 1.4V to take to 4 ghz. I have a C0 qx9650 that is stable with 1.37V, and I had a C1 Q9550 around for a while, that was stable at 3.4 ghz with 1.12V (OP says he needs 1.35V for 3.6 ghz!! wtfff). I really dont see how an E0 chip could be worse, from everything I have seen they are far better.

But the motherboard plays a big factor, especially when you disable vdroop, either by bios or by modding. you have to understand that once you disable vdroop, your VREG is 100% responsible for providing the most stable power possible. if you read the intel spec sheets, you will see that the Core 2 family is resiliant to voltage downshoot and overshoot, but only by 0.050V max, and 20 microsecond duration. You can push this envelope, but it could cause instability, so if your mosfets are shitty even with high voltages you will obtain shitty overclocks. You would need an oscilloscope to measure voltage under and overshoot magnitude and duration, under idle to load and load to idle state changes. You can however estimate at least magnitude by looking at how much your voltage drops under load with vdroop enabled. You also need to remember this magnitude is proportional with the amount of amps that go through the mosfets.

jjrowley,

VID is the voltage specified by intel that the CPU needs to run stable at stock voltage. So intel coded the cpu to request the bios for 1.25V at stock.

playing around with your cpu will not damage it. believe me, degradation, shortening life span, its all BS and placebo effect. what wears out is the other components, like the resistors, the mosfets, the capacitors. those wear out because when you overclock you are sending more amps through them than they are rated for.

 

[jjrowley]

so if I set 1.2500 as vcore I can leave everything else alone and figure max OC from that vcore and go from there.. square one... again!