Newbie Overclock Stumper

[Robot1501]

Really getting stumped... Hoping the gurus here can help me out.

I have recently had success overclocking my system from From 2.6 to 3.2 Ghz. (FSB from 333 to 400)

E6750
Asus P5KC (0701 Bios)
Crucial Ballistix (800Mhz 4,4,4,10)
CPU Voltage Auto
Stock Vdimm Auto
P182
Corsair 520 PSU
8800 GT
Zalman Fan

Core temps an a smidge above where they were at at stock speed. (speedfan and the crapy asus temp probe report about 13-22 Celcius at Load, with the cores in the low 30s.) It is rock solid stabe without any voltage increases and the temps are really low.

Now if I want to go higher than this (3.3GHz ,3.4 or 3.5) I start to run into problems... Even if I:

1. Increase voltage to 1.5
2. Increase VDIMM to 2.2
3. Relax the Ram timinings to 6-6-6-20

System loads up fine, but prime fails after a few minutes during torture test. I have a sneaky suspicion it is the RAM given that it usually fails during the RAM intensive portion of the test and the temps are all very low even when i bump up the voltage.

Is it likelly that my ram, which runs at very well at 800, 4,4,4,10 cannot even do 825,6,6,6,20?

I have heard elsewhere that the P5KC is finicky with the secondary timings and that asus default timing loads are quite aggresive. Could this be impacting things?

I am new to this, but i find it hard to understand how with no increase in voltages, i can overclock 20%, yet if I push voltages up and relax timings, I cannot even hit 30%. Would love some advice.

Cheers, and thanks to all who have contributed on this forum getting me to 20%

Robot

 

[iluvgillgill]

set the following voltage to
vcore 1.475V,
VVT/PLL to 1.5,
Northbridge 1.5V.
memory 5-5-5-15@800mhz@2.1V
2.2V is too high to start with.and you should know fsb any higher than 375 need a voltage increase on VVT/PLL and NB

 

[Robot1501]

set the following voltage to
vcore 1.475V,
VVT/PLL to 1.5,
Northbridge 1.5V.
memory 5-5-5-15@800mhz@2.1V


Does this mean that the memory will stay at 800 while i increase the FSB? (Asynchronous?)

I will give this a shot when i get home.. Thanks. I will let you know how it goes..

Thanks.

Robot1501

 

[iluvgillgill]

change the multiplier try to keep it 800mhz or under when you try overclock you cpu.select cpu multiplier if possible.higher fsb have better performance gain on memory speed.

 

[Evilonigiri]

Raising the NB, SB, and FSB voltage might solve your issue. It may be listed as some other thing like MCH voltage so look it up before changing them. Also, the ram voltage you set in the bios may be off a bit in real life. I set my voltage to 2V when it's really giving out 2.11V.

 

[Robot1501]

I tried what both of you suggested and still failures in prime 95...

I cannot find a multiplier that will let me keep the Ram at 400 (800mhz) or lower,while bumping the FSB above 400. (X8 for the current 3.2 GHZ). Is it true that the multipliers only work when you go "up" in the ram speed and not down?

I.E. I would like to run my ram at 667 or 800 while i push my FSB to 425, 450 or beyond... The P5KC bios does not seem to have this option.

Thanks again..

Robot1501

 

[Evilonigiri]

When you run the ram in sync with the fsb, the lowest you can do is 1:1, or ram multiplier of 2. Thus if you run 400fsb, the lowest you can do for ram is 800MHz.

 

[Robot1501]

Ok thanks Evil, I was not aware of that.

It looks like i am capped at 3.2 then. I am a bit supprised that i can push the ballistic 800 more than 820.

I either
A) Got some bum sticks
B) The P5KC is not the best for OC
C) Missed a tweak somewhere.... Dont think i did but will give it a shot.

Is it normal for ram to go from working great at stock voltages to not being able to run stress tests with max voltage in under 40 Mhz?

Robot..

 

[iluvgillgill]

well set your mem volt to 2.2 and loose your timing to 5-5-5-15.drop the multiplier to 7 see what you chan achieve.remember to raise voltage on other accordingly.

 

[Evilonigiri]

The Crucial Ballistic should be able to do 1000MHz. So 820MHz is nothing. I'm sure you can get up to 3.4GHz, but do you really need it?

 

[Robot1501]

Ok, thanks for the tip. I think your recommendation allowed me to rule out the ram as the issue...

I was able to get 450(FSB) X 7. The ram was effectivly hitting 900MHZ
When i try for 475, it wont even post..

I had
NB at 1.55
RAM at 2.2
CPU 1.475

It seems that my system can't press much more than 3.2... But now the blame lies on my CPU or the P5KC.

Ram is in the black slots.

Any other ideas guys?

 

[iluvgillgill]

up the nb to 1.6v+ did you also increase the VTT/PLL voltage?

 

[Robot1501]

I believe my MB goes from 1.55 to 1.7 (jumps past 1.6) . Is that voltage still safe?

As for the VTT\PLL

I only see one of them in my bios. Not sure what they are or what it stands for.

 

[iluvgillgill]

yeah.i listed 2 because different manufacture list it differently.just select the one near 1,6 if you go above 400fsb.increaseVTT(or PLL) which ever 1 is listed to 1.55V or similar.ry that.

 

[Robot1501]

Iluvgillgill, thanks for your suggestions and help, i have some succes in getting it to 3.4 and 3.5.

I am at a point now where i am happy with the outcome and do not want to push it anymore.

What did I do?

FSB 425
vcore 1.5
vdimm 2.1 5-5-5-16
VTT\PLL 1.55
Transaction Booster: DISABLED (This seemed to have an impact)
Relax level: 3
CPU Speed Spreader: Dis
PCI Speed spreader: Dis
ALL of the CPU options in the CPU section (SpeedStep, Execute Disable bit): Disabled

All in all, with the p5kc, i was able to push 3.2 at stock voltages and default bios settings but needed to tweak some others settings, (Trans Boost, CPU section) to get it stable...

I am going to try and put some of the other settings back and see which one seems to be causing the crashes at 3.4.

Would you feel comfortable with the settings above given that my Core temps are running around 45 degress at load (Using speedfan and asus) or 58 degrees at load (Core temp v0.96). Not sure why i am seeing a difference.. Does the core temp take the Tjunction into consideration and have a more accurate reading?

THanks again.

A happy P5KC overclocker...Robot1501

 

[iluvgillgill]

yeah that seems fine.but according to your info.you should able to achieve 3.5 or 3.6 without 1.5v on the vcore.you may have a low vid q6600.but i could be wrong.whats your vid listed in coretemp?what multiplier you using with 425fsb?

 

[Robot1501]

Hi Iluvgillgill

I actually have a e6750 (8x425) with a VID of 1.1875 in coretemp.

The Vcore is set at 1.5 and is usally around 1.44-148 in speedfan\asusprobe.

I reactivated all of the CPU fuctions tab in the bios (Speedstep, Execute disable bit etc..) and it still ran for 10 hours in prime with no errors.

I can conclude that one of the 3 variable below has allowed me to go from 3.2 to 3.4 as all of the other settings (Vcore 1.5 ,vtt\pll 1.55, vdimm 2.1, nb 1.5) have not changed.

Transaction Booster: DISABLED (This seemed to have an impact)
CPU Speed Spreader: Dis
PCI Speed spreader: Dis

Should i expect more?

Robot1501

 

[iluvgillgill]

sorry what i meant was the e6750.because i have been reply to too many q6600 thread so i just automatically type q6600
have you try lower the vcore after your success?

 

[Robot1501]

No worries. Your help has been great.

When i drop it down to 1.475, it crashes withing 10 mins.

Completed 3 hours of prime 95 @ 1.48 so it looks like i can bring the voltage down a very slight touch.

On another note, i have rolled the ram back down to 4-4-4-12 @ 425. How things change. What i initially thought was the ram, seems to not be the case at all.

The thing is, the heat seems well in check and i can push it to 3.6 GHz without much trouble yet it is not stable when torture test. Is there another setting i should be changing?

This overclocking bug is a deadly one when it gets you hooked.

Robot1501.

 

[Lupiron]

Dont we all know it!

--Lupi!

 

[Robot1501]

Does anyone have else have suggestions on why my Vcore needs to be so high (1.48) to make the OC stable? I find it very strange that I can OC at stock voltages to get 3.2, yet i need to go all the way to 1.48-1.5 just to hit 3.4 GHZ. Seems kinda surprising given that i have read you can normally inch the voltage upwards to get stability. I needed to take out the sledgehammer just to push it a tad above 3.2.

I see it as 1.46 (Vdroop t0 1.42) in speedfan. Would you gander the instability (failure of prime95) its a result of the Vdrop being so significant? IS their another setting i am missing?

PS I have posted the primary findings from this thread on a few other sites (asus forums and a few others) in hopes that other newbs who have had problems with the P5KC can learn from this. On another note. I am very pleased with this board, with the exception of its OC capability. If i would have known this, I probably would have elected for a P5K or a giga board.

ILuvgillgill thanks again.

 

[iluvgillgill]

but if your load temp is in safe range then not a worry.depend on the quality of your chip some do need the EXTRA boost in voltage to achieve.what's the setting on your memory if you are running on 3.6g?speed:timing@voltage

 

[Robot1501]

I had it running at 2.2, 5,5,5,15. WHen i couldnt push it any higher than that (to 3.6)i reduced it to 2.1, 4,4,4,12 to eek out some extra performance.

At load, it is about 1.48 according to speedfan when i run Prime95 (mixed torture test). Set to 1.5v in the bios. I wonder if i could push higher if i remove the dampner in the bios. From what i understand, this undervolts it a bit to minimize vdroop.

I could push it higher but from what i gather, anything over 1.5 is starting to get into dangerous territory, no? I have no issues if the CPU dies early, i will probably get a new quad in the next 24 months anyways. I just don't want it to go too early or risk damaging the MB or Ram, which i probably wont replace.

My concern is that the temps (58 degress at load) are not that accurate given what i have read.(Tjunction etc..) If i assume that 65 is the upper limit I am getting pretty close. I should probably start a poll on what people think is the maximum safe temperature for a e6750.

robot1501