• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Disadvantage of LLC?

Yuriman

Yuriman

Diamond Member
I've read in many places that LLC is bad for CPUs, that it causes voltage spikes that can damage a CPU if you're running a high voltage to begin with, etc., but what's the truth behind this for someone who's running a moderate overclock (on Ivy)? Is LLC good or not? Safe or unsafe?

I have an ASRock Z77 and a 3570K and I've noted a few things about running with LLC vs without. Voltage fluctuations are much more visible with LLC disabled (or rather at level 5, the lowest setting).

I was running happily at 4.5ghz with a small bump in voltage over stock (15mv), but after reading a lot of negative opinion I decided to try and get my same overclock stable without LLC. With LLC on, when loading the CPU, the voltage as reported by CPU-Z would scale smoothly with CPU load and level out at 1.280v. When the CPU load is reduced, the voltage drops back down. I went into bios and disabled LLC and upon reboot, immediately crashed. I discovered that my voltage without LLC was dropping to about 1.21v, which is obviously far too low for this overclock. I bumped my voltage up until I had compensated enough to get back up to 1.280v (+115mv) under load, but noticed that when I only have two cores loaded, or when I'm unloading the CPU (and right before it clocks down) the voltage will jump up to something like 1.32v. So what it *appears* LLC is doing is preventing voltage jumps caused by vdroop going away when under partial load.

I can't see how LLC could be bad in this light. I'll do some further testing though and see if I'm able to get stable at a lower voltage without LLC on, but what I've seen so far suggests that conventional wisdom about LLC is backward.

Thoughts?
 
Last edited:
LLC is safe as long as you moderate your voltage. Running 1.51v with LLC on full is crazy, but 1.4v with LLC on would be fine. It just reduces teh effect of Vdrop that 9for no apparent reason) motherboard manufacturers implement into the VRM's
 
acchceheheheemmm...

Ok....

Load Line calibration TODAY, is better than it was only a few years back, Most of the platforms out now "SINCE, P67" by asus/gigabyte/asrock handle voltages much more quickly and precisely than the p5q days.


LLC used to be a problem because it attempted to lock voltages to a certain setting, and this caused voltage spikes that overshoots the target voltage significantly.

This was also only visible on an Oscilloscope. Not in any monitoring software.

TODAY, however, this does not happen, the vrms respond much faster and more precisely.

Load Line Calibration, now.., only serves to scale voltages evenly across processor loads independent of the vdroop protection built into Intel specifications.
 
borisvodofsky said:
acchceheheheemmm...

Ok....

Load Line calibration TODAY, is better than it was only a few years back, Most of the platforms out now "SINCE, P67" by asus/gigabyte/asrock handle voltages much more quickly and precisely than the p5q days.


LLC used to be a problem because it attempted to lock voltages to a certain setting, and this caused voltage spikes that overshoots the target voltage significantly.

This was also only visible on an Oscilloscope. Not in any monitoring software.

TODAY, however, this does not happen, the vrms respond much faster and more precisely.

Load Line Calibration, now.., only serves to scale voltages evenly across processor loads independent of the vdroop protection built into Intel specifications.
Click to expand...

How does analog (asrock) vs digital (asus) VRM's effect the current implementation of LLC?
 
Many will be shocked to find out what LLC does to vCore in reality, by measuring vCore with a digital multimeter.

I try to avoid using LLC as much as possible.
 
lopri said:
Many will be shocked to find out what LLC does to vCore in reality, by measuring vCore with a digital multimeter.

I try to avoid using LLC as much as possible.
Click to expand...

Is this the modern implementation of LLC, or is this the LLC that was used back in the AthlonX/P4 days?
 
LLC is safe people.. Use it.

Even before the current generation of improved LLC, the high voltage spikes didn't cause any problems except at extremely high overclocks.

These were the guys who thought they were safe behind their water cooled fortresses of solitude.

Either way, even if degradation occurs, it's a very gradual process, takes years and years and years.

PC are good for 3 years top before being obsolete. And by that time, if they can't handle 4.8ghz anymore, just drop down to 4.5, I guarantee it'll still hold that.
 
Last edited:
borisvodofsky said:
LLC is safe people.. Use it.

Even before the current generation of improved LLC, the high voltage spikes didn't cause any problems except at extremely high overclocks.

These were the guys who thought they were safe behind their water cooled fortresses of solitude.

Either way, even if degradation occurs, it's a very gradual process, takes years and years and years.

PC are good for 3 years top before being obsolete. And by that time, if they can't handle 4.8ghz anymore, just drop down to 4.5, I guarantee it'll still hold that.
Click to expand...

Sounds like a lot of baseless guess work going on here.

You say it's safe then say degradation may occur in 3 years... What are you basing it's safety on and where are you getting these numbers? Much less guaranteeing them? I've read articles about LLC and it's negative side affects, I haven't ready anything saying "hey, it's safe to use now" and if there is such an article saying so, I'd like a link to read it and enlighten myself.
 
2is said:
Sounds like a lot of baseless guess work going on here.

You say it's safe then say degradation may occur in 3 years... What are you basing it's safety on and where are you getting these numbers? Much less guaranteeing them? I've read articles about LLC and it's negative side affects, I haven't ready anything saying "hey, it's safe to use now" and if there is such an article saying so, I'd like a link to read it and enlighten myself.
Click to expand...

http://www.overclockers.com/load-line-calibration/

TAAA DAAAAAA 😎😎😎😎
 
How about an article where the author himself doesn't concede that his scope may have too slow a refresh to catch the spikes?
 
Another question: can the voltage reported by CPU-Z be trusted? I seem to be stable at a lower reported voltage now than I needed with LLC enabled, but I'm suspicious of CPU-Z's numbers.
 
Yuriman said:
Another question: can the voltage reported by CPU-Z be trusted? I seem to be stable at a lower reported voltage now than I needed with LLC enabled, but I'm suspicious of CPU-Z's numbers.
Click to expand...

No CPUz numbers are scaled. I can't seem to find that equation to "descale" it.

But it's around the forums somewhere.

LLC is safe. PERIOD
 
2is said:
How about an article where the author himself doesn't concede that his scope may have too slow a refresh to catch the spikes?
Click to expand...

Nope, he did not concede, he reset his settings to max, and ran the test multiple times. No spikes to be seen no where.

and THIS IS ON p5q, very old stuff. 😱

You're just wrong about everything tonight, not your night, //pat pat.
 
borisvodofsky said:
Nope, he did not concede, he reset his settings to max, and ran the test multiple times. No spikes to be seen no where.

and THIS IS ON p5q, very old stuff. 😱

You're just wrong about everything tonight, not your night, //pat pat.
Click to expand...

It is also possible that my scope’s 50 microsecond refresh is too slow to capture the spike but I don’t think that this is likely myself.
Click to expand...

🙄🙄
 
2is said:
Sure do. Do you know what "too slow" means? Don't get upset with me, this is coming from YOUR source. 😉
Click to expand...

You clearly do not have an understanding of Oscilloscopes.

quote author:

"Inspected the scope carefully, for those who care:
50uS per screen (50 microseconds), the finest point it can detect/display appears to be 90-120 nanoseconds, somewhere in that area."
🙄🙄

Since the tests are have no clear start/finish, 90ns is more than enough to capture the spikes even without staggering

I'd seriously come over there and rub ur head from all the ownage i've been handing you this morning. 😀😀
 
Last edited:
I quoted the author directly too except in my quote he says in plain English it may not be fast enough. You're just quoting numbers and applying your own guess work by claiming its more than enough.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top