45nm Intel CPU Failure Resulting From Loadline Calibration Enabled

Check some of the other threads.
It's already been shown that it is MORE dangerous to raise idle voltage to compensate for vdroop's dips, than to have loadline/vdroop mod at a lower constant idle/load voltage), and have spikes to what your idle would be at the higher voltage (without vdroop). Pretty obvious actually.
1.35 idle/1.35 load with spikes up to 1.42v when switching states is safer than 1.42 idle/1.35 load, since the chip is exposed to less of the higher voltages. (Note that the average computer will spend more time idle than load, unless you're a cruncher or power off the computer as soon as you're done with any game or other load activity).

This is board dependent though. Proper working boards should have LLC no higher than bios setting or maybe .01v higher. If a board overvolts past bios setting at load, with LLC, in some cases, a lot more past cpu-z readings (taking with a DMM) then that's not normal and that's very bad.

any thoughts, opinions or experiences with LLC on i7's ???

i really havnt been using it but I am creepiong up slowly and finding max OC at stock volts and having fun with it :p

Quote:

---

Originally Posted by Rattle

any thoughts, opinions or experiences with LLC on i7's ???

i really havnt been using it but I am creepiong up slowly and finding max OC at stock volts and having fun with it :p

---

There never was a problem with LLC and Core 2. It is fine with i7.

I have always run my oc'ed systems with LLC enabled 24/7, P5K-P with Q9450, Maximus II Formula with Q9650 and P6T-D with i7 920. Operates as it should on all boards, no problems at all with any system to date. Maybe they will show degredation around 2015 or 2020 or perhaps they won't. If they do I'll let you know, if not then that will be the end of it. :ROTF:

I overclocked an i7 920 C0 at 4515 Mhz under phase, LLC enabled on a Gigabyte X58-UD4P motherboard and I noticed that my CPU was not stable any longer under Prime95, after a few weeks. The cause may be the voltages since I applied 1.35 V for the uncore (QPI/VTT) and 1.72 V on RAM. The other voltages were not high :

CPU : 1.475 V (bios)
CPU PLL : 1.8 V
QPI PLL : 1.2 V
IOH : 1.28 V
ICH : 1.14 V

Then I sold it and bought a 920 D0 3845B : the same phenomenon happened whereas every voltage was very low, always under phase : first fully stable at 4661 Mhz under OCCT and Prime 95 after 2 hours, it was no longer stable after a few weeks. In this case, the voltages were the following :

CPU : 1.375 V (bios)? LLC enabled
CPU PLL : 1.8 V
QPI/VTT : 1.22 v
QPI PLL : 1.1 V
IOH : 1.1 V
ICH I/O : 1.5 v
ICH : 1.1 V
RAM : 1.58 V

For this second CPU, I really wonder about the degradation cause : it is neither the temperature since the CPU never exceeded -8°C, nor the voltages since most of them are near or at the minimum level.

May the cause be LLC ?

I already wonder how you can run such OC's with such low QPI volts... did you try other rams ? Did you need a higher Vcore to stabilise now ?

For the 920C0, the uncore frequency was at 3440 Mhz and the RAM, at 1720. So the QPI voltage had to be at 1.35 V.

For the D0, the uncore frequency was 3108 Mhz and the RAM at 1333 Mhz. So, only 1.22 V was required.

Moreover, the extreme cold enables low QPI voltages.

With the 920 C0, I tried 3x2 GB GSkill Pi-series and it was impossible to obtain a stable overclock, the stability behaviour was random : sometimes, Memtest was fully stable 3 times and after a reboot, Memtest did not pass once.

I did not test with a higher Vcore up to now since I changed the motherboard and I did not mount the evaporator head yet. So I am running stock frequency, on air, during a few weeks.

Well the thing about LLC and people saying it's dangerous having the vcore jump up.
Take my board for example (R2E), having LLC off that means my vdroop at 1.35v idle/bios~ drops to almost 1.31~ during full load, then when going off load it boosts up to 1.35 again. But with LLC enabled it goes up to 1.37~ instead during full load, then back to 1.35 on idle. So I actually have less volt variation with LLC turned on than off.
To run my computer stable at 4.4, with LLC off, I would need to pump about 1.41 at idle to maintain a 1.37~ at load. Which would be 0.6 more volt on idle than i need now. Which honestly in my opinion is far worse and "damaging" on the CPU than 1.35 idle and 1.37 load.
This might just be my board and bad vdroop though, but it seems logic to me that the volt changes with LLC on is less "damaging" than with off.

Edit: Seems like I had vcore jump and overvolt spikes mixed up as Leeghoofd pointed out. Anyhow, I never noticed any odd spikes with the dmm future on my RIIE during prime. Maybe need to sample a bit longer.

It's not the Vcore up jump , it's the overvoltage spikes (at eg 1.4Vcore set, that ya CPU sometimes can get 1.6 or such during milliseconds)

LLC is best put off when benching (strain on the mobo's PWM's) for daily use I don't see an issue with it... unless there's something seriously wrong with ya board...

I couldn't seem to get my 4GHz o/c stable when LLC was disabled, but it was no problem stablizing it with it enabled..

I will keep it enabled on my systems.

Even the andantech article waveforms showed the same proportional voltage overshoot with LLC enabled or disabled, as would make the most sense.

If I have LLC off, bios volts = max overshoot volts. So assuming board is following intel specs:
If I set 1.35v in bios, my idle vdrops to 1.3v (so overshoots from idle never go above 1.35) and load will be 1.28 (so overshoots from load never go above 1.35). So overshoots would be .07v max at load, assuming the bios appropriately calibrated vdroop + vdrop such that what you set in bios = maximum overshoot.

If I then set loadline at 1.28v, because at load that is what I need to be stable, and lets assume LLC was perfect and no vdrop/vdroop. Overshoot still occurs .07 over load or to 1.35v, so nothing changed, except at idle I am now at 1.28v instead of 1.35v, but both settings has same temporary overshoots.

I think as long as you realize, ie in this example, 1.28 LLC on = same overshoot as 1.35 LLC off, and not for example assuming 1.45vcore LLC on = 1.45 vcore LLC off, I dont see the issue.

Until someone with a 5K+ oscilliscope can prove that overshoots are worse proportionately with LLC, I will be using LLC, and actually even that would not stop me at this point, since i have been using LLC so long, I would then conclude any difference is apparently not harmful given my particular cpu use.

No problems here with my E8400E0.
Gamed 24/7 at 4.55Ghz @1.480v with LLC on for 6 months, served me well till the day I took it out to make place for a Q9550E0 :D
Its a keeper :yepp:

From my understanding using LLC in conjunction with EIST or C1E can cause dangerous voltage spikes. This is when core clock and/or voltage/FSB is consently changing of the CPU cycling with C1E or EIST. So it's best to use LLC when EIST and C1E is disabled.

Quote:

---

Originally Posted by BenchZowner

Grabs the Mythbusting stick and posts:

First we had the High PLL, or high VTT ( FSB Termination ) killed our chips and we're sure and blah blah...
I also remember the "deadly" voltages in a table in the review of the Asus Striker II Extreme.
And then... a lunatic decides to test lots of CPUs under crazy conditions to see if they're degrading or even dieing when using high voltages ( and sometimes... insanely high voltages )... bang
And then... Core 2 Goes To Hollywood... 1.6Vcore LoadLine Calibration Enabled 2.0Vtt ( FSB Termination ) 2.8Vpll running F@H 24/7 @ Full load... lives on for days and days without a simple sign of degradation, nor death.

As for LLC itself... I've been running all my systems and CPUs with no Vdroop ( either Vmod or LLC enabled where applicable ) since the first Conroe E.S. appeared... that's more than 2 full years, with not a single dead CPU from LLC or high voltages... and everybody on these forums knows what kind of voltages I run no matter what cooling I'm using :D

---

+1 to that. I remember the crazy guy that clocked a ton of CPU's way past 1.8vcore with LLC.

As for the scardie cats, don't use LLC then. No harm, no foul right?

I use LLC and it gives me higher OC, less heat, and very little voltage "spike" (it's more like a dribble). If some people are going to call me a noob for that then so be it. Yes I understand the risks and the concerns, but that doesn't mean I respect them.

On the flip side I know one person that fried an e8500 with some pretty low voltage and LLC happened to be enabled. Just because LLC happened to be enabled people jump to the conclusion that it killed the chip. Fact is, we don't have enough details. What I do know is that I've been running it and benching for days on end. My chip is fine and like I said the voltage spikes and temperatures are all in check. After all, the max spec for my 1156 chip is 1.55vcore and @ 1.39vcore and LLC enabled I'm not even close. Max it ever spikes to is 1.40v.

I use Asus Rampage 2 Gene CPU 1.192 it overvolts from time to time to 1.2 Volts which isn't bad at all. No stability issues, but i might do some tests, what program do you use to monitor the voltages?

Are you guys certain that we shouldn't be running LLC with EIST and C1E?

LLC allows me to drive the voltage from 1.4v to 1.375v (reads in CPU-Z as 1.344 v)

EIST and C1E are nice when the CPU clocks down to 6x multiplier (2.27 GHz) and drops vcore to 1.050v.

So yes at the moment I am running all three. Is this a bad idea? I'm even running all of this with Gigabyte's dynamic energy saver which is throttling the CPU and managing power usage and saving watts somehow. Tells me I've saved 97 Watts in the past 5 hours. I don't know where it's getting it's energy savings from.

Am I doing any damage to my CPU with all of this?

Quote:

---

Originally Posted by mikeyakame

Just slightly above what the real vCore is, if you get 1.41v idle/load by sensors, aim for around 10mV-20mV (0.01 - 0.02v) above the actual vCore. I know the P45 boards don;'t have vTT monitoring which makes determining it a little harder but if you are using LLC I would set the vTT to around 1.38 - 1.40v max and assume a slight overvoltage due to LLC which is safe to say. Otherwise if no LLC enabled set it around 1.44v. The Rampage Formula overshoots by about 40-50mV with LLC enabled. It's safer to assume this is true when you can't monitor it via sensors. vPLL would probably want to be around 1.60-1.64V. vPLL has a very easy method to gauge whether it's right or not with Prime95. All you need to do is monitor the difference between both threads for either core and how long apart they take to complete each loop. Dual core makes it a little tougher since OS related usage can effect the accuracy, but close what you don't need running while Priming and you shouldn't get too much error in ur results from outside system resource use.

---

Awesome thanks mikeyakame, time to get to work :hammer:

When I did a degree in electronics my Lecturer (Prof or Dr in electronics), said that he leaves his PC on as its safer than turning it on and off. Ever pushed the light switch only for the bulb to blow? Its the voltage transients (spikes) that are damaging regardless of if they breech the max vcore limits.

i also had a PSU die when I flicked the wall switch with the PSU switch on, now I always turn one thing on at a time.

What is this?! People talking about scary thing called overvolting and LLC?! Am I still @ xtremesystems, or this is some other forum? :confused:

Faulty components cause blowouts. If no components are faulty there is exactly the same risk in constantly leaving your system on as there is turning it off and on. Its not necessarily voltage that kills either, but the current. Higher voltage = higher current.

But the problem is only with high voltages right?

Example on i7, voltages greater than 1.4V (1.55V specifications)

my question is, when set to auto is it on or off? hmmmm

Quote:

---

Originally Posted by Chrysalis

my question is, when set to auto is it on or off? hmmmm

---

easy

if u don't see vrood from idle to full is enabled

however on new motherboards there are more step of llc

on my p55a ud6 there are 3 step

off, intel specification
level 1, moderate vdroop
level 2, no vdroop