Q6600 overclocking problem

WingZero30 · 21 May 2011 at 09:44

Hi guys

I recently managed to overclock Q6600 to 3.5GHz

. Bios vcore is 1.4v and load voltage is 1.36v. It is IBT stable.

However I decided to push it further to 3.6GHz.

I set the bios vcore to 1.425v and load voltage was 1.392-1.376v (IIRC). I ran IBT and on the 3rd pass IBT failed giving the usual message of better cooling or reducing overclock.

So I increased vcore to 1.45v giving load voltage of 1.408v in cpu-z. However on the first pass of IBT the whole PC crashed.

I think it is either the voltage regulator of mobo (4 phase only), northbridge cooling or the PSU itself which can't supply enough current.

I made a thread couple of weeks ago about northbridge getting too hot as it is passively cooled. I can only place a finger on it's heatsink for 1 second.

BTw mobo is Gigabyte GA EP45 UD3LR:

http://www.gigabyte.com/products/product-page.aspx?pid=2951#ov

PSU is OCZ modxstream 600W which has been brilliant so far:

http://www.overclockersclub.com/reviews/ocz_mxs_pro_600w/

Ram is Kingston Hyper X DDR2 1066MHz:

http://www.overclockers.co.uk/showproduct.php?prodid=MY-039-KS

mraerosmith1981 · 21 May 2011 at 09:50

Grab an Antec Spotcool, and have it blow on the NB

WingZero30 · 23 May 2011 at 12:20

My [email protected] is Prime95 stable (small FFTs stress test). I am using latest version V26.6 64 bit. It seems to have a new SUM(INPUT) error checking which I also checked along with round off error.

LLC enabled.

Bios VCore: 1.4375V

Idle Voltage: 1.408V

Load Voltage: 1.392V

DRAM speed: 1066MHz

DRAM voltage: 2.00V

My idle temps are higher than usual because I was testing Prime 95 at different speeds so didn't shut down the PC to let it cool for some time.

Plus the room was closed.

However the thing is my motherboard can't handle the IBT stress testing for [email protected] and beyond. It usually fails on 3rd pass. I increased vcore to 1.45V and then to 1.475V giving load voltage of 1.44V but still the pattern is same.

At these voltages I can hear grinding squeeking noises coming from my motherboard. I suspect this may be due to my voltage regulator system being stressed out which is only 4 phase plus as stated before northbridge cooling is very poor.
So motherboard can't handle such settings and it crashes.

I think I will need the 16 phase Asus P5Q Deluxe which also has a very nice chipset cooling for overclocking my Q6600 further.

Mick-1965 · 23 May 2011 at 13:18

Almost identical problem when I upgraded my sons machine to a q6600, except his mother board was a P5B ( 3 phase ) and I could only achieve 3.2
The chipset would quite easily handle 400 bus as it'd been there before with a C2D, except now with the quad the power regulator MOSFET's were scalding hot under load ( IBT ).

Picked up a vanilla P5Q (8phase) off of feebay and It'll quite happily do 3.6 now

As you say the Q6600 demands allot of power when over clocked and a quality multi phase power regulation is a must.

WingZero30 · 24 May 2011 at 23:12

I am still trying to stabilise my [email protected] (400 MHz x 9) with IBT. LLC enabled.

Basically [email protected] (389MHz x 9) is IBT rock stable.

Bios Vcore 1.40V
Load voltage 1.36V

However I have noticed very strange phenomena with my motherboard Gigabyte GA EP45 UD3LR for [email protected]:

Further I increase Bios Vcore beyond 1.40V, quicker the IBT crashes!!

So if bios vcore is 1.41v - 1.43v, my rig can pass 2 -3 runs of IBT before it crashes

If bios vcore is 1.43-1.45v, my rig can only pass 1 run of |IBT before it crashes

If bios vcore is 1.45v - 1.475v, my rig crashes after few seconds

Another peculiar thing I have noticed is shown below.

Bios vcore is 1.4125V
Load voltage is 1.376V

Now I have disabled both C1E and EIST in bios, so my multi should be staying at x9 throughout the testing with core speed at 3.6Ghz. However just before posting output results, the multi decreases to x6 giving core speed of 2.4GHz!!. Hence 44 and 43 GFlops. This seems to happen when the system is under immense stress.

Taking into account Prime95 testing during which Bios Vcore was 1.4365V I have come to following conclusions :

My motherboard/rig can handle higher voltages when load isn't extreme (Prime 95)

My motherboard/rig can't handle higher voltages when load is extreme (IBT)

It can although handle extreme loads at lower voltages.

Seems to me that 4 phase system and especially chokes/mosfets can't handle increased voltages.

Or may be PSU can't supply enough current/voltage although it is OCZ Modxstream 600W with two 12V rails each capable of supplying 25amps. So shouldn't that be 12V x 25A = 300W on each rail?

Or the current from wall socket isn't enough?

Anyone can shed further light on this issue or did any one else had same problem?

For now I can only say that [email protected] is IBT stable and [email protected] is Prime95 stable.

My motherboard just can't seem to hack it :mad:

unless I am missing something.

Any help will be greatly appreciated :cool:

setter · 24 May 2011 at 23:31

When i had my old q6600, (in a p5q deluxe), at 3.6ghz on 1.4 vcore, i had to do a lot of tweaking of NB, PLL and FSB - TERM voltages, a good high phase board certainly will help, looks like a decent vid chip you have there as well, mine was 1.2750 vid, eventually got it to 3.8ghz.

WingZero30 · 24 May 2011 at 23:35

setter said:
When i had my old q6600, (in a p5q deluxe), at 3.6ghz on 1.4 vcore, i had to do a lot of tweaking of NB, PLL and FSB - TERM voltages, a good high phase board certainly will help, looks like a decent vid chip you have there as well, mine was 1.2750 vid, eventually got it to 3.8ghz.

Would you still have those settings with you to give some rough idea?

I am going notch by nocth of bios vcore until I can see where I face serious problem then start changing some other values like PLL, VTT etc.

WingZero30 · 25 May 2011 at 08:38

I was testing by increasing bios vcore one notch at a time as stated. I increased the;

Bios vcore : 1.43125V
Idle Voltage : 1.408V
Load Voltage: 1.392V

IBT was running fine until again on the 3rd run, the multi dropped from x9 to x6 slowing down the test and giving different GFlops value.

So I restart my PC, went into bios and increased vcore further to;

Bios Vcore: 1.43750v
Idle Voltage : 1.408V
Load Voltage: 1.392V

Also I decided to change the following voltages from Auto to Default values except for MCH (Northbridge) voltage as shown :

CPU Termination: 1.200V
CPU PLL: 1.550V
CPU Reference: 0.805V
MCH Core: 1.100V (increased to 1.400V)
MCH Reference: 0.760V

Started IBT again and after 1st run, my PC crashed again

.

After crash my PC tried to reboot but couldn't. Everytime it restarted, it would try to reboot but would turn off again. It kept repeating it until I manually turned off my PC. Turned off from the back. Gave it some break and turned on again.

Luckily it turned on fine this time and I went into bios and got the follwoing message:

The system has experienced boot failures because of overclocking or changes in voltage. Latest settings in BIOS my not coincide with current H/W states

So for now I have gone back to [email protected].

Any further suggestions?

Mick-1965 · 25 May 2011 at 09:28

You are at the limits of what your VRM's can produce in terms of power ( wattage ).

In laymen's terms, around the outside of the CPU socket there are 4 grey silvery cubes ( 1 for each phase), next to these you'll find 3 small chips. These chips control the voltage and current going to the CPU.
You'll probably find that these are scalding hot when under load. you could help by giving these extra cooling.

WingZero30 · 25 May 2011 at 09:51

Mick-1965 said:
You are at the limits of what your VRM's can produce in terms of power ( wattage ).

In laymen's terms, around the outside of the CPU socket there are 4 grey silvery cubes ( 1 for each phase), next to these you'll find 3 small chips. These chips control the voltage and current going to the CPU.
You'll probably find that these are scalding hot when under load. you could help by giving these extra cooling.

Yeah those small black chips are mosfets next to a big black square which is a choke.

I also think that my VRM has reached it's limit of producing 'clean' required voltage to CPU. I think by testing at IBT level at those GFlops values, I am overloading my 4 phase VRM which it can't handle and just crashes.

It's a shame I am mobo limited with my Q6600 overclocking potential.

Diggsy · 25 May 2011 at 09:59

My Q6600 is exactly the same - Prime stable for hours on end, but not IBT stable. I just gave up as nothing I use will ever hit it like IBT. As far as I am concerned, if it's Prime stable then its fine

WingZero30 · 25 May 2011 at 10:09

Diggsy said:
My Q6600 is exactly the same - Prime stable for hours on end, but not IBT stable. I just gave up as nothing I use will ever hit it like IBT. As far as I am concerned, if it's Prime stable then its fine

Yeah agree

.

My Q6600 is a good overclocker and can run at high speeds with lower voltages. Only requires 1.36V for 3.5Ghz and that is for IBT alone.

I just retested with Prim95 again yesterday and it was again rock stable after 10 hours and 13 mins for [email protected]. This time I increased bios vcore slightly more to 1.45V.

Mick-1965 · 25 May 2011 at 10:18

WingZero30 said:
Yeah those small black chips are mosfets next to a big black square which is a choke.

I also think that my VRM has reached it's limit of producing 'clean' required voltage to CPU. I think by testing at IBT level at those GFlops values, I am overloading my 4 phase VRM which it can't handle and just crashes.

It's a shame I am mobo limited with my Q6600 overclocking potential.

Yeah I know what everything is, just wasn't sure that you did, hence the laymen explanation.

If you have any small heat sinks that'll fit and some thermal epoxy to glue them on with, it may help.

or without going to the extreme of gluing them on, just for test purposes. Lay you case on its side, so that the board is flat. mount them with a small amount of heatsink compound.

WingZero30 · 25 May 2011 at 10:34

Mick-1965 said:
I know what everything is

Mick-1965 said:
If you have any small heat sinks that'll fit and some thermal epoxy to glue them on with, it may help.

or without going to the extreme of gluing them on, just for test purposes. Lay you case on its side, so that the board is flat. mount them with a small amount of heatsink compound.

That's a good tip. So the thinking here is that if I place thermal compound on mosfets, it will hopefully help by transferring the heat away from the surface of mosfets.

Should I also place some on top of chokes aswell?

Have you ever done anything like this before?

WingZero30 · 25 May 2011 at 11:06

Now I know why Asus P5Q Deluxe motherboard rocks. It's cooling system is highly efficient with heatsinks also placed on mosfets.

I think next time when I build a new PC I will keep the following things in mind aswell when choosing new motherboard:

Power phase VRM design

Cooling solution on chipsets and mosfets

Comparing Asus P5Q Deluxe and my Gigabyte GA EP45 UD3LR

Asus P5Q Deluxe

-16 Phase VRM
-Very nice cooling solution for chipsets
-mosfets also have heatsink

Gigabyte GA EP45 UD3LR

-4 Phase VRM
-Just a basic heatsink on chipsets
-bare mosfets

No wonder people can overclock their [email protected] and beyond on Asus P5Q Deluxe. It is afterall widely regarded as the best skt 775 motherboard.

Mick-1965 · 25 May 2011 at 11:17

WingZero30 said:
That's a good tip. So the thinking here is that if I place thermal compound on mosfets, it will hopefully help by transferring the heat away from the surface of mosfets.

Should I also place some on top of chokes aswell?

Have you ever done anything like this before?

No, I mean just use compound to hold the heat sinks in place while you are testing. Assuming the case is on its side with the board flat.
I don't think putting compound on the mosfets on its own would help any if at all.
Just temporary option while you can decide if heat sinks help or not. If they do help then you can glue them on with thermal epoxy.

WingZero30 · 25 May 2011 at 11:31

So far I can only see arctic silver thermal epoxy.

Is there mx-4 or mx-3 thermal epoxy aswell?

Mick-1965 · 25 May 2011 at 12:13

WingZero30 said:
So far I can only see arctic silver thermal epoxy.

Is there mx-4 or mx-3 thermal epoxy aswell?

The only thermal epoxy ive ever used is artic silver.

Are you not going to test first ?

WingZero30 · 25 May 2011 at 12:50

Mick-1965 said:
The only thermal epoxy ive ever used is artic silver.

Are you not going to test first ?

LMAO

I don't even have small heatsinks. I am trying to remove small heatsinks from my brother's dud old graphics card. Those heatsinks have thermal pads though. Don't know if they can be reattached. I have to see.

Mick-1965 · 25 May 2011 at 13:51

WingZero30 said:
LMAO

I don't even have small heatsinks. I am trying to remove small heatsinks from my brother's dud old graphics card. Those heatsinks have thermal pads though. Don't know if they can be reattached. I have to see.

As above, lye the case on its side and just use a blob of thermal compound to hold the heat sink in place temporarily just while you test. No need to re-use the pads. As long as you're not shaking the machine around or try to stand it up they shouldn't go any where.
I had a heatsink on the back of a graphics card for a few weeks mounted this way with arctic ceramique. As I had no thermal epoxy left at the time