LLC on Gigabyte UD5 on or off......

Soldato
Joined
12 Jul 2009
Posts
4,878
Curious as to what other people are setting Load line calibration to, enabled or disabled in bios.

Some say its dangerous to enable it due to spikes in voltage when going from idle to load and vice versa

others say not.....

What to do...?
 
Personally if overclocking to high voltages on the cpu it's best to be off.

so is it ok if the voltages are low to begin with, I have a i7 920 D0, which is stable with 1.25v in bios and LLC enabled, but needs 1.287v with it Disabled.

Maximum vcore for I7 is 1.375 according to intel.
 
I'm with greebo on this one. Turned off, because I'm likely to put high voltages through the chip.

Extensive discussion here

LLC lets you seelct lower idle voltages, at the cost of stressing the motherboard and passing transient high voltage spikes through the processor. Setting lower idle voltages just doesn't matter, so I'm not going to stress the board and the cpu more to achieve this.
 
Last edited:
I have never really tested my system with LLC disabled, I have always been under the impression that having LLC enabled improves stability when overclocking, so I always have it turned on...
 
It seems people who are using 45nm quads are disabling it, yet people with i7 cpus are enabling it. i7 D0 are using far lower volts than the quads anyway. the question is are the i7's spiking within the parameters of the cpu's or not with LLC enabled as most are using far less volts than the intel specs allow.

so if I disable Load Line Calibration and set my vcore to 1.287v that is the absolute maximum that the voltage can go..? in other words, any voltage spike will be below 1.287v. But if I enable LLC and set a lower voltage, lets say 1.25v then the voltage spike can go above this figure say to say 1.4v or higher, is this correct?.

What about voltage regulation from the motherboard itself, should the motherboard ensure a steady voltage and compensate for this. Also if LLC is so damaging to 45nm CPU's why on earth is it on every x58 motherboard. Why doesnt the motherboard regulate the transients in the first place.

I'm not an electrical engineer, but my next door neighbour is and he says that if you are using low voltages to begin with, then the transients shouldnt break the upper limit of the cpu in the first place. All the info I have found relates to 45nm quads using high voltages in the first place. How does this relate to an i7 with low volts.?
 
Last edited:
Then your next door neighbour doesn't realise just how quickly a processor goes from idle to load, and so just how high the spikes are. Show him the graph from anand.

It's bad for i7 for exactly the same reason it was bad for the quads, except that the issue is less with low voltages and a lot of people on the i7 seem to be temperature limited and not voltage.

Everything I can find on the topic points at people new to the game and people who just haven't researched it all have it turned on, and the more experienced people turn it off.

@bradley, generally I consider anything over the vid range for the processor to be high voltages. So for the i7, anything over 1.375V is 'high'.

I really need to spend some time working through the implications of blacks equation to arrive at a 'sane' higher voltage to use, I'm trying to offset decrease in temperatures against increase in voltages. Something along the lines of '90 degrees at 1.25V volts is considered fine, so how high can the volts be at 50 degrees'. It shouldn't be hard mathematically, I just haven't got round to it yet.

Probably staying with above ambient temperatures for the forseeable future, though I'm still looking at peltiers. Here's hoping a 240 and two 120s is more than enough to keep an i7 cold. May be lapping, unsure
 
Did you try increasing the vcore to the point where under load it is at the same voltage as with llc on when under load? If not I'm unsurprised it crashed

Morrowind?
 
Back
Top Bottom