Welcome Guest ( Log In | Register )

Outline · [ Standard ] · Linear+

 Guyz do you enable L.L.C. for 24/7 use ?, L.L.C. => Load Line Calibration

views
     
lohwenli
post Apr 19 2009, 10:09 PM

Penang Overclockers Club
*****
Senior Member
971 posts

Joined: Aug 2006
From: Penang


QUOTE(monsh @ Apr 10 2009, 02:44 PM)
is it same as advanced clock calibration ?
*
No, AAC is quite different. Its got something to do with the clock pulse adjustment.

QUOTE(bryanyeo87 @ Apr 10 2009, 04:36 PM)
i dont use it, i use a pencil on the buck pwn controller, much much more effective brows.gif
*
There are risks in this, see below.

QUOTE(mercury @ Apr 10 2009, 05:08 PM)
Yea, they do use different names, thx for the info about what issit being named by diff manufacturer.
Erm, yea. Just curious, does AMD Procs have vdroops like Intel's ?
*
Basically what this thread is talking about is vDroop compensation. What must be understood is that vdroop is actually a safety design in motherboard voltage regulators; when powering a power hungry processor, the vCore is delibrately allowed to drop, especially under load. The reason is that when the processor load changes, the voltage will fluctuate, and may in fact overshoot the target voltage.

When a processor shifts from low load to high load, the voltage will actually drop briefly before the voltage regulator compensates, this is why overclockers don't like vdroop-it may drop voltages to below acceptable levels.

The problem is when a processor shifts from high to low load. When the power consumption drops, the regulator is still pumping in lots of current causing voltage to spike until the regulator compensates. The higher the change in power consumption, the worse the spike. By allowing voltage to drop under load, the severity of the spike can be minimized. This voltage spike can be a greater danger to the processor than setting high voltages to compensate for vDroop, and unless you have a high speed recording oscilloscope you'll have no way of checking (voltage spike will be shorter than 0.1s, but that's enough to cause damage).

I found out all this from a processor article on Penryn, but I can't remember the site or I'd list it here. Its a lot easier to understand with the graphs in the original article.

 

Change to:
| Lo-Fi Version
0.0420sec    0.65    7 queries    GZIP Disabled
Time is now: 9th December 2025 - 09:07 PM