• 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?

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I'm not sure what you mean by spectrum, but yea there is more than one full load, that's the problem.

Intel VID is what's screwing people over.

A better overclocking feature would be for the motherboard to determine voltage directly

So you can set a voltage for EVERY load condition and multiplier separately. :sneaky:

Yea, we'd have hundreds of value to test, but that's awesome. 😎

I mean any load states below full load/full speed where vdroop isn't a factor.
 
But if you want to be a smartass, try not to confuse physics with math. They aren't interchangeable.

Did you go to one of those special schools where they teach you about voltage in maths class? Oh and if you want to be a smartass try learning to read.
 
Did you go to one of those special schools where they teach you about voltage in maths class? Oh and if you want to be a smartass try learning to read.

No but we aren't actually discussing the concept of voltage, but the differences in the numbers behind them. That's math. Try again.

I also read just fine, just a simple misunderstanding as to what was being referenced.

I'm not the one being a smart ass, you're the one trying to be and failing at it. Very successful at acting like a douche though. Bravo to you.
 
Last edited:
No but we aren't actually discussing the concept of voltage, but the differences in the numbers behind them. That's math. Try again.

I also read just fine, just a simple misunderstanding as to what was being referenced.

I'm not the one being a smart ass, you're the one trying to be and failing at it. Very successful at acting like a douche though. Bravo to you.

Troll battle?

As I recall I won both our previous engagements. 😀


I still don't understand how the VID is determined though. I can't find a graph or chart /table of the VID conditions.

I know it's different for each CPU, but which conditions are there that sets each step.
 
Troll battle?

As I recall I won both our previous engagements. 😀


I still don't understand how the VID is determined though. I can't find a graph or chart /table of the VID conditions.

I know it's different for each CPU, but which conditions are there that sets each step.

Nah, you quit remember. 😉
 
I was under the assumption that offset in bios skews the entire voltage range, while "additional turbo voltage" in Intel's Extreme Tuning Utility only changes voltage in turbo frequencies.
 
No but we aren't actually discussing the concept of voltage, but the differences in the numbers behind them. That's math. Try again.

I also read just fine, just a simple misunderstanding as to what was being referenced.

I'm not the one being a smart ass, you're the one trying to be and failing at it. Very successful at acting like a douche though. Bravo to you.

You read something wrong, I pointed your mistake out, you claimed you didn't make a mistake and started trying to teach me something I already know, I pointed out the exact mistake you made and made a smartass comment because you seem to think you are infallible, you made a smartass comment back because that is your only defence at this point, I made another smartass comment, you tried to twist the arguement to suit your needs......

The upshot of the whole thing is.... you were still wrong, admit it and let it go you are boring me.

P.s Bravo to you too, I will leave you to argue with Boris now he seems to have more spare time than I do.
 
You read something wrong, I pointed your mistake out, you claimed you didn't make a mistake and started trying to teach me something I already know, I pointed out the exact mistake you made and made a smartass comment because you seem to think you are infallible, you made a smartass comment back because that is your only defence at this point, I made another smartass comment, you tried to twist the arguement to suit your needs......

The upshot of the whole thing is.... you were still wrong, admit it and let it go you are boring me.

P.s Bravo to you too, I will leave you to argue with Boris now he seems to have more spare time than I do.

I think you're just reading wrong. I never commented on the actual voltage figures, just the effect of LLC and I already said I had misunderstood what the OP was refferring to when he said lower voltage.

You bolded the section you're referring to, I bolded the section I was commenting on. Go back and read it, they weren't the same thing. If you're getting bored and feel like you've made your case, then stop responding.

P.S. Boris and I are merely reminiscing.
 
Last edited:
I'm not sure what you mean by spectrum, but yea there is more than one full load, that's the problem.

Intel VID is what's screwing people over.

A better overclocking feature would be for the motherboard to determine voltage directly

So you can set a voltage for EVERY load condition and multiplier separately. :sneaky:

Yea, we'd have hundreds of value to test, but that's awesome. 😎


Umm we have that, it's called speedstep..........

And LLC is a perfect way to make your system unstable. Just because you stress tested doesn't mean you're stable.

Want to make your computer crash? Use something like Linx, and use a tiny problem size (so that it finishes very quickly). Let it run for a while, and your computer will BSOD, guaranteed.

Under shoots are completely uncontrollable with LLC, and eventually you'll dip just below your lowest stable voltage.
 
Umm we have that, it's called speedstep..........

And LLC is a perfect way to make your system unstable. Just because you stress tested doesn't mean you're stable.

Want to make your computer crash? Use something like Linx, and use a tiny problem size (so that it finishes very quickly). Let it run for a while, and your computer will BSOD, guaranteed.

Under shoots are completely uncontrollable with LLC, and eventually you'll dip just below your lowest stable voltage.

Which is interesting, because the original argument against LLC was because it caused overshoots.

My experience has suggested that it's more likely to cause undershoots now, which means I have to increase my voltage to get stable.

Has the implementation of LLC changed?
 
User adjustable LLC is a hack by motherboard makers to counteract a problem which doesn't exist. Somebody decided vdroop was an issue. Intel certainly didn't decide this, because vdroop is actually in their CPU specifications.

Answer: LLC is not needed on a motherboard that meets Intel specifications. You don't even have the means of measuring the need or the implementation.

If you need to use LLC, you have other problems. Period.
 
Last edited:
User adjustable LLC is a hack by motherboard makers to counteract a problem which doesn't exist. Somebody decided vdroop was an issue. Intel certainly didn't decide this, because vdroop is actually in their CPU specifications.

Answer: LLC is not needed on a motherboard that meets Intel specifications. You don't even have the means of measuring the need or the implementation.

If you need to use LLC, you have other problems. Period.

LLC has nothing to do with combating V-droop

The point of LLC is to scale the voltages properly, and have the minimum required voltages at the Highest load.


Because the VID has too many steps for a SINGLE offset function to cater to the entire range of possible loads.


You have no idea what you're talking about. 😀
 
For a given vrm ciruitry/quantity & quality of output caps, overshoots and undershoots will occur of the SAME magnitude whether LLC is ON or OFF. Only the peak to peak transient is different since vdroop (active voltage positioning) lowers the voltage at load, so the overshoot that occurs goes back closer to baseline, and the undershoot prior doesnt have as far to recover. Vdroop (active voltage positioning) is simply a cost effective means of reducing output caps, while maintaining a specified peak to peak transient.

If your concerned about magnitude of overshoots and undershoots that is only helped by better vrm circuitry/more output caps, so you should be complaining to those buying cheaper mobos. Cheaper vrm/cheaper mobo allows larger magnitude overshoots and undershoots, but then uses more active voltage regulation (higher vdroop) to contain them to spec peak to peak.

Then if you want to complain LLC is bad based on 25-30% higher peak to peak transients, then, again, first you need to go around and tell people to quit overclocking with cheap motherboards since they will have worse specs yet (even though we all know they work fine for OCing). High end motherboard, like GB UD5 or Asus top, etc, that have many times number of output caps would maintain both lower magnitude under and overshoots and a smaller peak to peak even with LLC on moderate level, than a cheaper boards using magnitude lower output caps and full Vdroop, ONCE you start overclocking. Yet people dont seem to have an issue overclocking with cheap mobos.

That being said, I would not use LLC so high that vcore is higher at load than idle, that makes little sense, and can lead to instability at idle. But using LLC to decease vdroop to a small amount .02-.04v or so on a decent mobo with many output caps makes more sense than not to me, and mathematically irrefutable to still have lower magnitude overshoots, undershoots, and peak to peak transients than OCing with cheaper mobos. In grand scheme, all you are doing with a decent mobo, is lowering your idle vcore by using LLC.

As an aside, you need 1ghz bandwidth capable of 1-5 nanoseconds updating to reliably capture transient peaks (overshoots and undershoots).

But bottom line, degradation isnt a matter of LLC on or off, it is a matter of what voltage your cpu sees, however you get there.
 
Last edited:
And therefore there's no need to adjust LLC.

Thank you.

LLC gives you a smoother voltage curve with varying loads. That is what most people here are using it for. 😀

IT is absolutely necessary. Because without it, your peak load voltage will be much higher than necessary.

There is a large VID difference between FULL LOAD, with and without the memory controller engaged.

Without memory controller engaged, like in Prime 95 small ftt, the CPU only asks for 1.38v-1.39v at 4.8ghz

With the memory controller engaged, such as IBT or Prime custom using all ram, the CPU will ask for 1.41v-1.42v

That is a huge difference.

If you don't smooth the curve with LLC, you're be looking at 1.3v without engaging memory controller, and 1.36v with memory controller.

The 1.3v @ 4.8ghz will crash, while the 1.36v is fine, so you'll fail prime small ftt, but pass intel burn test.

NOW Scenario 2, WITH LLC

I get 1.33v without engaging the memory controller and 1.35v with memory controller engaged.

This is a more ideal situation. 😎


Vdroop has nothing to do with stability in my case.
 
AGAIN LLC is not used to "combat" vdroop. It INFLUENCES Vdroop, but vdroop isn't the problem.

The problem it solves, is the VOLTAGE GRADIENT issue.
 
Umm we have that, it's called speedstep..........

And LLC is a perfect way to make your system unstable. Just because you stress tested doesn't mean you're stable.

Want to make your computer crash? Use something like Linx, and use a tiny problem size (so that it finishes very quickly). Let it run for a while, and your computer will BSOD, guaranteed.

Under shoots are completely uncontrollable with LLC, and eventually you'll dip just below your lowest stable voltage.

Nah...I'll tell you with LLC enabled I am 100% guaranteed stable at a specific voltage offset and with LLC off I am also 100% guaranteed stable at another offset and it doesn't matter what test you try, it will still remain stable with zero WHEA errors if I have the correct offset set in the bios.
 
^^, I agree, those mythical undershoots, do NOT happen often, if at all.

The most important thing is that LLC gives me a voltage gradient that is better acclimated to my processor load.
 
Nah...I'll tell you with LLC enabled I am 100% guaranteed stable at a specific voltage offset and with LLC off I am also 100% guaranteed stable at another offset and it doesn't matter what test you try, it will still remain stable with zero WHEA errors if I have the correct offset set in the bios.

Ill have to agree as well. Mainly because I just tried it. Took the smallest sample size and keyed in 1000 passes. Not only did it not BSOD but it passed.
 
Nah...I'll tell you with LLC enabled I am 100% guaranteed stable at a specific voltage offset and with LLC off I am also 100% guaranteed stable at another offset and it doesn't matter what test you try, it will still remain stable with zero WHEA errors if I have the correct offset set in the bios.


Yea you think you are, but you're not.

Think of vdroop as shocks in your car. It's there to help smooth out the potholes and speed bumps. It doesn't mean you can't drive without shocks, you sure can, but eventually you'll break your car.

Same thing here. Eventually you'll hit a big enough voltage transient to make your system crash. And you'll never reproduce this while stress testing because it happens randomly during idle/load transitions. It's going to be unpredictable, unreproducible, and uncontrollable.

If you like running you're system like that, by all means go ahead. If vdroop wasn't necessary Intel engineers would have already removed it.
 
Back
Top