Load line calibration not working

[Carfax]

I don't think LLC is working on my Asus P5K.

My CPU voltage is set to 1.4 in the BIOS, and my Intel Q9450 is clocked at 3.6ghz (450x8).

It's rock solid stable at that clockspeed, but if I try to go over 450 by even 1 mhz, it always crashes during prime.

Even though Anandtech recommended leaving LLC alone, I enabled it figuring it would help to keep my PC maintain stability at a FSB over 450 by reducing V-droop.

So I set the FSB to 455 and boot up into windows and start up Prime95, while monitoring the voltage with CPUZ.

Idle voltage is 1.36v, but when the processor is under full load, the voltage drops to 1.28/9 and my computer freezes.

I was under the impression that LLC was supposed to reduce V-droop, but on my motherboard, it seems to do nothing.

Or maybe I'm not understanding it clearly....

Any opinions?

[BenchZowner]

As far as I remember LLC wasn't working on he non-deluxe P5K ( your motherboard, that is ).
At 450MHz is the voltage display fluctuating in CPUz ?

Have you used a DMM ( Digital MultiMeter ) to measure the real voltage supplied to your CPU ?

[Carfax]

No I haven't used a DMM, but if LLC purportedly doesn't work on the non deluxe P5K, then that definitely explains my issues

I actually wanted the Deluxe version as well, but Microcenter never had it in stock. Thats what I get for supporting my local computer store rather than buying from Newegg

I guess it doesn't really matter though. I was just trying to see if I could break the 450 FSB barrier on this set up. Insofar however, all of my efforts have been for naught.

At 450MHz is the voltage display fluctuating in CPUz ?

You mean under load? No, the voltage drops from idle 1.36 to 1.28 during load and remains more or less constant with very little fluctuation.

[BenchZowner]

Your instability is more than likely caused by the FSB itself and not your CPU's voltage.
You should drop your CPU's multiplier to 6x ( CPU Ratio ) and work your way up ( upping the FSB ) and testing for stability.
You might hit a wall ( already ) or just need to tweak your BIOS settings ( FSB Termination Voltage, Memory/Chipset timings, PLL Voltage & GTL's if available )

[Carfax]

FSB termination voltage is set to 1.4

Memory is set to 2.2v

Chipset is set to 1.5

PLL... Hmm, I think I have that on auto. How useful is PLL for getting a higher FSB? (*edit* nevermind, I just checked. It's at 1.5v)

GTL ratio is set to .67.

I'll lower the multiplier like you suggested though, and see if I can run a higher FSB. I really want to crack this 450 barrier

[BenchZowner]

To maximize your max stable FSB you'll have to play with various FSB Termination Voltage & GTL Ratio combinations.
For example at 460MHz FSB your CPU might be stable with 1.34Vfsb and 0.61x CPU GTL, and at 470MHz with 1.4Vfsb and 0.63x CPU GTL.

[Xvys]

I wonder why your voltages are so high? I know quads take a bit more voltage than dual cores, but some o/c'ers have better luck with lower voltages. I would check out the numerous threads about Q6600 and the 45nm Quads o/c'ing for some tips...If I remember correctly there seems to be a FSB wall around 450 with the Quads.

I have the P5K Deluxe and the LLC does work. I think it keeps the actual voltage closer to what you set in the bios. vDroop under load is another matter. My P5K has a consistent .008v drop under full load as compared idle voltages. No matter what I set the cpu voltage to, I always get a .008v vDroop. I always have LLC enabled. I wonder if your power supply is being overtaxed? I have a 600w Fortron.

With my E3110 @ 4100 MHz (513 x 8), I set 1.3875v in bios which shows as 1.368v at idle, and 1.36v at full load. Here are my voltage settings which are 100% stable:

Asus P5K-D/E3100

cpu: 1.3875v
ram: 1.9v @ 1026 MHz - 5-5-5-12 timings
VTT: 1.1v
PLL: 1.5v
NB: 1.25v
SB: 1.05v
cpu GTL: auto
nb GTL: .61x

Bootstrap: 400 MHz
Loadline Calibration: enabled
Overcharge voltage: disabled
Slews: auto
Transaction Booster: enabled, "2" boost = PL9 (I would set this to disabled while testing)

[Carfax]

To maximize your max stable FSB you'll have to play with various FSB Termination Voltage & GTL Ratio combinations.
For example at 460MHz FSB your CPU might be stable with 1.34Vfsb and 0.61x CPU GTL, and at 470MHz with 1.4Vfsb and 0.63x CPU GTL.

These are good suggestions, I'll try them thank you

[Carfax]

I wonder why your voltages are so high? I know quads take a bit more voltage than dual cores, but some o/c'ers have better luck with lower voltages. I would check out the numerous threads about Q6600 and the 45nm Quads o/c'ing for some tips...If I remember correctly there seems to be a FSB wall around 450 with the Quads.

My voltage may seem high, but it really isn't.

Me setting it at 1.4 in the BIOS, actually turns out to be 1.36v in Windows.

Also, under load, the processor consumes only 1.29v according to CPUz.

This was why I wanted to try LLC, because as soon as the processor starts gunning up, the voltage droops big time.

I wonder if your power supply is being overtaxed? I have a 600w Fortron.

I don't think it's the power supply. I have a 750 watt Thermaltake.

With my E3110 @ 4100 MHz (513 x 8), I set 1.3875v in bios which shows as 1.368v at idle, and 1.36v at full load. Here are my voltage settings which are 100% stable:

See, due to LLC, your load voltage isn't much lower than what you set it in the BIOS. On the regular P5K though, the voltage drops precipitously as soon as the processor goes under load.