• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Vdroop? Do you leave it enabled?

B

BababooeyHTJ

Supreme [H]ardness
Joined
Jan 21, 2009
Messages
6,951
We have all seen the Anand article and have seen some of the discussions on the forums about this topic. Do you guys disable vdroop? I have been using it on my last few builds but am considering disabling it for my current build.
 
I think you mean Loadline Calibration, which is designed to combat Vdroop. You can't disable Vdroop itself. Most everyone is using LLC for the SB chips, at least.
 
I always enable Loadline Calibration because i want the closest vcore to the one i set in BIOS.
 
Forceman said:
I think you mean Loadline Calibration, which is designed to combat Vdroop. You can't disable Vdroop itself. Most everyone is using LLC for the SB chips, at least.
Click to expand...

With Asus and Gigabyte that is what it is called. Not with MSI and Biostar, IIRRC. :rolleyes:
 
Well, you can't just "turn off" vdroop, that's for sure. You can turn off the BIOS features that help reduce it, that's not a very good idea though.
 
I run my i7 930 at 4.3Ghz with LLC off. I set 1.375v in bios and CPUz reads 1.35v.
 
I leave LLC on for my i7-870, i5-750 and i7-950 builds. I like the idea of getting more life out of cpus.
 
CrayMan said:
I leave LLC on for my i7-870, i5-750 and i7-950 builds. I like the idea of getting more life out of cpus.
Click to expand...
LLC will technically reduce the life of your CPUs while overclocking in exchange for keeping the average voltage closer to the level you set it to.
 
Yeah I decided against it on my current board. Enabling LLC caused the voltage under load to go up which means I needed more voltage than necessary under load to keep the cpu at idle speeds stable. That caused power consumption to go up quite a bit and temps to skyrocket.

I don't remember it being so bad on my UD3.

Zero82z said:
LLC will technically reduce the life of your CPUs while overclocking in exchange for keeping the average voltage closer to the level you set it to.
Click to expand...

It's the wattage that kills not necessarily just the voltage. Your cpu doesn't draw much while idle.
 
BababooeyHTJ said:
It's the wattage that kills not necessarily just the voltage. Your cpu doesn't draw much while idle.
Click to expand...
Actually, it is the voltage that kills. CPUs are semiconductor devices, and applying too much voltage will damage them.
 
Zero82z said:
Actually, it is the voltage that kills. CPUs are semiconductor devices, and applying too much voltage will damage them.
Click to expand...

Not necessarily, more voltage seems to equal more wattage. Thats why I have a feeling that a higher voltage under load is worse than a higher voltage while idle.

I'm not talking about obscene voltages like 1.5v or anything like that. Yeah, that can kill a cpu.
 
BababooeyHTJ said:
Not necessarily, more voltage seems to equal more wattage. Thats why I have a feeling that a higher voltage under load is worse than a higher voltage while idle.
Click to expand...
More voltage does mean more power, but that is not a danger to the CPU. It doesn't really matter if the high voltage is while the CPU is idle or at load. What matters is whether or not the voltage is within safe limits for the CPU. If it is higher than that, then it will cause damage.
 
Zero82z said:
LLC will technically reduce the life of your CPUs while overclocking in exchange for keeping the average voltage closer to the level you set it to.
Click to expand...

Zero82z said:
More voltage does mean more power, but that is not a danger to the CPU. It doesn't really matter if the high voltage is while the CPU is idle or at load. What matters is whether or not the voltage is within safe limits for the CPU. If it is higher than that, then it will cause damage.
Click to expand...

:confused:
 
86 5.0L said:
LLC will overcompensate as a CPU down clocks and has the potential to shoot past the safe limit.
Click to expand...

and that is more dangerous than a slightly higher idle voltage how? What is this theoretical safe point and how does it relate to LLC? I think that the higher voltage under load that I seem to need to use with LLC on my current board is the riskyer part.
 
BababooeyHTJ said:
and that is more dangerous than a slightly higher idle voltage how? I think that the higher voltage under load that I seem to need to use with LLC on my current board is the riskyer part.
Click to expand...

a high voltage that is under the limit is safer then a sudden spike over the limit
 
BababooeyHTJ said:
and that is more dangerous than a slightly higher idle voltage how? What is this theoretical safe point and how does it relate to LLC? I think that the higher voltage under load that I seem to need to use with LLC on my current board is the riskyer part.
Click to expand...

86 5.0L said:
a high voltage that is under the limit is safer then a sudden spike over the limit
Click to expand...
^ Bingo.
 
Yes I can also confirm that turning LLC on may decrease CPU life more because it will spike when it tries to compensate for the vdrop.

Voltage is what kills CPUs. The more voltage you give it the faster the transistors turn on and off. The transistors only have a certain limit of how many times they can handle being toggled and will eventually just wear out.
 
SirMaster said:
Voltage is what kills CPUs. The more voltage you give it the faster the transistors turn on and off. The transistors only have a certain limit of how many times they can handle being toggled and will eventually just wear out.
Click to expand...

Yeah, that's not the cause. Voltage has nothing to do with how fast the transistors toggle - that's what the frequency controls. The problem with high voltage is electromigration - which leads to intermittent errors and finally failure.
 
What is this theoretical limit that you guys keep referring to? How much of a difference in vcc measured under load between having LLC on or off is enough to trigger this limit?
 
The maximum safe voltage depends on the CPU. I don't know if it's really possible to know how much of a voltage spike there is while the CPU transitions from idle to load states without actually measuring it with an oscilloscope.
 
Zero82z said:
The maximum safe voltage depends on the CPU. I don't know if it's really possible to know how much of a voltage spike there is while the CPU transitions from idle to load states without actually measuring it with an oscilloscope.
Click to expand...

So how are you jumping to these conclusions? Much like with the whole VTT scare Anandtech isn't above jumping to conclusions. I don't remember seeing an osilloscope mentioned in the Anandtech article.

I just find it hard to believe that a small spike in voltage with a very small load (with very little current) above some theoretical limit would be the factor that causes a cpu to degrade over time but a slightly lower voltage under load (with real current) being drawn would be no issue.
 
How am I jumping to conclusions? Enabling LLC will cause the CPU to be exposed to higher voltage spikes than it would normally be if Vdroop was allowed. That is a fact. Whether or not those voltages are dangerous is not something most people can say, but there's no doubt that the risk of damage is higher.

Whether or not you think that higher power consumption is more damaging than higher voltages is irrelevant. This isn't about what you think, it's about the properties of CPUs and other semiconductor devices. Higher voltages lead to more electromigration.
 
Zero82z said:
How am I jumping to conclusions? Enabling LLC will cause the CPU to be exposed to higher voltage spikes than it would normally be if Vdroop was allowed. That is a fact. Whether or not those voltages are dangerous is not something most people can say, but there's no doubt that the risk of damage is higher.

Whether or not you think that higher power consumption is more damaging than higher voltages is irrelevant. This isn't about what you think, it's about the properties of CPUs and other semiconductor devices. Higher voltages lead to more electromigration.
Click to expand...

No this is what you have been arguing. Voltage is also not the only factor in electromigration.

I just mentioned one thing that I had noticed when enabling LLC on my board and you jump in and basically say "no, no thats wrong" and repeat the Anand article verbatim.
 
Forceman said:
The problem with high voltage is electromigration - which leads to intermittent errors and finally failure.
Click to expand...

yeah but is that even a factor? does it reduce a CPUs life span from 10 years to 9 years? 10 to 1?

my i7 920 has been folding fine for 1.5 years at 100*C and 1.35v+ still holds a 200 bclk :cool:
 
BababooeyHTJ said:
No this is what you have been arguing. Voltage is also not the only factor in electromigration.

I just mentioned one thing that I had noticed when enabling LLC on my board and you jump in and basically say "no, no thats wrong" and repeat the Anand article verbatim.
Click to expand...
Temperature is also a factor in electromigration. However, with LLC disabled, the higher voltage will occur when the CPU is at idle, so the temperature will be quite low, especially if the CPU's power-saving features are enabled (which will decrease both frequency and voltage).

You didn't mention something you noticed. You replied directly to one of my posts and said that high power consumption and not voltage is the main factor that leads to CPU damage. That is not true. And yes, I'm basically repeating what the Anandtech article said, because that article is correct.
86 5.0L said:
yeah but is that even a factor? does it reduce a CPUs life span from 10 years to 9 years? 10 to 1?

my i7 920 has been folding fine for 1.5 years at 100*C and 1.35v+ still holds a 200 bclk :cool:
Click to expand...
Electromigration is certainly a concern, especially at higher voltages. 1.35V isn't a problem for a 45nm i7, but if you look at people who do extreme overclocking and run their CPUs at extremely high voltages (1.6V and up, let's say), they actually kill quite a few CPUs.
 
I've gone upto 1.45v before, I cant even imagine 1.6v being healthy for it lol.
 
86 5.0L said:
yeah but is that even a factor? does it reduce a CPUs life span from 10 years to 9 years? 10 to 1?

my i7 920 has been folding fine for 1.5 years at 100*C and 1.35v+ still holds a 200 bclk :cool:
Click to expand...

That's the $300 question. No one knows for sure, besides maybe Intel, what the safe voltage really is - and even if you exceed the safe voltage you still might not kill the CPU, a lot depends on luck. Which is why I don't see using LLC as a problem - the random chance of the voltage spikes being big enough (and lucky enough) to kill the CPU is worth the trade off of having a stable load Vcore, at least for me.
 
Forceman said:
That's the $300 question. No one knows for sure, besides maybe Intel, what the safe voltage really is - and even if you exceed the safe voltage you still might not kill the CPU, a lot depends on luck. Which is why I don't see using LLC as a problem - the random chance of the voltage spikes being big enough (and lucky enough) to kill the CPU is worth the trade off of having a stable load Vcore, at least for me.
Click to expand...
I don't see LLC as a problem either, providing you're using reasonable voltages to begin with. However, it is important to understand that there is technically an extra risk when using it. When you're overclocking, the more understanding you have of the factors at play, the better off you are.
jojo69 said:
Anyone considering load line calibration should read and understand http://www.anandtech.com/show/2404/5 scary stuff
Click to expand...
That's what we've been discussing for most of the thread...
 
Forceman said:
Yeah, that's not the cause. Voltage has nothing to do with how fast the transistors toggle - that's what the frequency controls. The problem with high voltage is electromigration - which leads to intermittent errors and finally failure.
Click to expand...

I'm not an expert in this at all but I was just quoting an Intel Engineeer in this respect.

"Voltage kills our processors"

"Heat will never kill our processors. It hits a certain temperature, it throttles, goes a little bit higher, shuts down."

"If we increase the voltage above what our spec is, we're moving those transistors back and forth faster and you start shortening the life of those transistors."

"The minute you overclock it, the life starts going down"

"By keeping the voltage down, you don't bring the life down as fast"

Video of the engineer:
http://www.intel.com/Assets/Video/r...width=340&height=270&title=Unlocked Advantage
 
86 5.0L said:
a high voltage that is under the limit is safer then a sudden spike over the limit
Click to expand...

Undershoot and Overshoot
Google this ISL6336

DROOP allows the regulator to operate as expected with reduced output capicitance allowing manufacturers to reduce the number of output caps. Disabling droop is beneficial to overclocking if you can ensure stability while not giving excess voltage to the processor, but does require more output capacitors than with droop enabled.
Foxconn and DFi do not follow the Intel recommended loadline like the other manufacturers do.
 
Guys I just wanted to offer up a logical idea about Vdroop. I don't know much about how it works, only that it's a relatively new technology. Some argue that Vdroop has the potential to cause harm instead of good. As a new technology, if it was more trouble than it was worth I don't understand why leading motherboard manufacturers like ASUS, MSI, Gigabyte, etc. would implement a technology with side-affects that could cause significant harm. Maybe my idea is naive and places too much trust in the engineers, but I doubt they would implement a technology that does more harm than good, or any significant damage for that matter.

I'll continue reading about Vdroop, your responses and the AT article ^^^ got me interested. Thanks guys.
 
XacTactX said:
Guys I just wanted to offer up a logical idea about Vdroop. I don't know much about how it works, only that it's a relatively new technology. Some argue that Vdroop has the potential to cause harm instead of good. As a new technology, if it was more trouble than it was worth I don't understand why leading motherboard manufacturers like ASUS, MSI, Gigabyte, etc. would implement a technology with side-affects that could cause significant harm. Maybe my idea is naive and places too much trust in the engineers, but I doubt they would implement a technology that does more harm than good, or any significant damage for that matter.

I'll continue reading about Vdroop, your responses and the AT article ^^^ got me interested. Thanks guys.
Click to expand...
I think you mean load line calibration, not Vdroop. Anyway, if you leave your CPU at stock voltage, LLC won't harm anything. That's probably also true for reasonable voltage increases during overclocking. However, as always, whenever you change the motherboard settings to run your components at higher speeds and voltages than they are specified with, you run the risk of damaging something, and your hardware isn't covered by warranty any more.
 
Zero82z said:
That's what we've been discussing for most of the thread...
Click to expand...

Yeah, I know, but a lot of folks are just skimming...it bears repeating. I read a lot out there where people think that LLC is some magic bullet. That overshoot is the moment where your CPU goes away.
 
XacTactX said:
Guys I just wanted to offer up a logical idea about Vdroop. I don't know much about how it works, only that it's a relatively new technology. Some argue that Vdroop has the potential to cause harm instead of good. As a new technology, if it was more trouble than it was worth I don't understand why leading motherboard manufacturers like ASUS, MSI, Gigabyte, etc. would implement a technology with side-affects that could cause significant harm. Maybe my idea is naive and places too much trust in the engineers, but I doubt they would implement a technology that does more harm than good, or any significant damage for that matter.

I'll continue reading about Vdroop, your responses and the AT article ^^^ got me interested. Thanks guys.
Click to expand...

its implemented by intel and its a cheap way to make mobos by skimming on the output caps.
 
jojo69 said:
Yeah, I know, but a lot of folks are just skimming...it bears repeating. I read a lot out there where people think that LLC is some magic bullet. That overshoot is the moment where your CPU goes away.
Click to expand...

depends on the quality of your board and the total of output caps.

Btw saw some one mention a osiliscope up there somewhere lol

The only way to test it is with a Intel Voltage Regulator Test tool aka VRTT
 
You must log in or register to reply here.
Back
Top