Another Q6600 & IP35 Pro OC thread - Help needed

GoogalyMoogaly

GoogalyMoogaly

GoogalyMoogaly

Soldato
Joined
29 Aug 2010
Posts
8,620
Location
Cornwall
Hi,
I know there have probably been more threads on this than just about anything else on OcUK, but that's just making it harder for me to find the right thread!

Basically I'm pretty new to OCing and could do with some pointers with regards to OCing this combo.

I get the FSB and multiplier stuff, but it's the voltages that really get me confused.
Are there any threads that give these settings? I've found various threads with them in but it's usually people saying they don't work.

I'm looking get get a 3.0GHz overclock for starters, something that will hopefully run cool and pretty quiet. After that I'd like to go for 3.2 and/or 3.4GHz. I'm not too fussed about 3.6GHz but if I can and it's not too hot then it'd be nice.

I'm running an ABit IP35 Pro with a Q6600 (G0) and 4GB (2 x 2GB) Corsair XMS2.
For cooling I'm using a simple WC loop consisting of XSPC X20 750 pump/res, XSPC Delta V3 CPU block and a 240mm EK CoolStream rad with 2 x 1850rpm Gentle Typhoons (@ ~1100rpm currently). With the exception of the fans, these are my spare WC parts so I know they're not great but figured they should comparable to a half decent air-cooling solution right?

It's in a CM 690 II case with the top side panel fan as an intake to try to cool the mobo (since there's not HSF to move air around that section).

I'm currently using:
Ext clock - 333
Multiplier - 9x
DRAM - 1:1.25 (DDR2-799)
CPU Core- 1.3050v
DDR2 - Auto
CPU VTT - 1.23V
MCH 1.25v - 1.29v
ICH - 1.05v
ICHIO - 1.50V

and one core keeps seeming to hang when running Prime95, it doesn't fail but it just stops. The other seem to keep going though and none of the temps seem to be an issue (CPU 49ºC, System 30ºC, PWM 66ºC).

Any pointers as to which thread I should be looking at?

Thanks.
 
Last edited:
Take DDR2 off auto, and set the voltage manually.

Up the NB by a couple of notches
Set the voltage on the VCore to at least 1.35V (I know you can go lower, but this will show that it can pass prime etc if it works).
Don't worry about the other voltages at the moment, at this level of overclock they don't really come into play.

BTW.. I 'think' MCH is the northbridge.

Also.. some people use 8x multiplier and get better results.. it's a chip to chip thing.
 
I decided to try running IBT.
At stock I can run this using the Standard setting and it seems to pass (I think, well, all the results come back the same). However if I try Custom and up the memory then it fails almost instantly.
Considering when I was running Prime I was using the Blend test, could this indicate a memory issue?

I have run MemTest+ v2.20 (or whatever the newest one is) though and it didn't report any errors.
 
you are right in thinking you are not stable.
have you tried the settings Scougar suggested?

also when running IBT open task manager and look in the performance tab and see what is the 'free' ram amount.
enter just below this in IBT, eg, i have 2600mb free so i enter 2500.
set threads for 4 and run 10 passes, if the test runs fine, then imo you are stable.

if the ram is at its stock speed, then it should be fine, but it may be worth removing a stick of ram and making a record of the ram timings and voltage, printed on the sticker

have a look at the ram speed in the bios, it should say 800mhz etc.... or its stock speed, and see what its voltage is, does it match whats on the stick?

My DFI mobo has horrible ram vdroop and is off by .1v so i had to raise it by this.
also check that the ram timings match whats on the stick (my timings are left on auto and they do match)

also my q6600 did not like the x9 multi above 3.0ghz, i tried it with a x8 multi and it was much better. so try 375 x 8 = 3000mhz

my B3 revision needs 1.44v to run at 3.4ghz (425 x 8)
 
I've seem to post this all the time now, hopefully it helps you out, I know it has helped a lot of others out...

My settings for 3.6 (my chip is a 1.3 VID)...

External Clock - 400
Multiplier - 9x
DRAM Speed - 1:1:25 (DDR2-1000)
PCI-E - 100

CPU - 1.52v (1.42 or so under load)
DDR2 - 2.0v
CPU VTT - 1.23v
MCH - 1.29v
ICH - 1.05v
ICHIO - 1.5v
DDR2 REF - 0%
CPU GTL 0+2 - 67%
CPU GTL 1+3 - 67%

Limit CPUID MaxVal - Disabled
C1E - Auto
Execute Disable Bit - Enabled
Virtualisation - Disabled
EIST - Auto

RAM - By SPD

...you will need to watch the PWM temps. My board is bolt modded and I have a 60mm fan hovering above the PWM area keeping them cool. You may well need the bolt mod if it's not already done... http://www.xtremesystems.org/forums/showpost.php?p=2330347&postcount=751

I also have 2x 40mm fans on the Northbridge, as it can get pretty toasty too.

Bear in mind that the above setting was when I was running 4gb RAM. I am still at 3.6 with 8gb, but have to run the memory on a different divider so that it runs at 800mhz. Any more was too much for the board to cope with.

EDIT - for what it's worth I found that anything above 1.3 for the MCH would cause instabilities. For some reason my board prefered the MCH to be kept fairly low.
Click to expand...

Given that you are watercooled, then I would suggest that cooling the PWM's with some air will be very important, as they will have no air moving over them at all and will very quickly get VERY hot.

Oh and before I forget, with regards the RAM voltage, bear in mind these boards tend to overvolt it just a tad, so don't go too crazy here. I would simply set it to the manufacturers spec, or 1 notch lower.

And another thing, make sure you haven't set the memory timings aggressively, as this will almost guarantee you failures.
 
Last edited:
j.col said:
also when running IBT open task manager and look in the performance tab and see what is the 'free' ram amount.
enter just below this in IBT, eg, i have 2600mb free so i enter 2500.
set threads for 4 and run 10 passes, if the test runs fine, then imo you are stable.
Click to expand...

It says under "Physical Memory" (using WinXP) Available = 255xxxx (<-- last bit keeps changing)

So in IBT I enter 2400 and it says:
You cannot use more than 2047 MB of memory to test, because you have a 32-bit OS. If you wish to test using more memory, you'll need to upgrade to a 64-bit OS. The test will now start using a memory size of 2047 MB.
Click to expand...
Apart from odd comma placement (imo) that seems to make sense.

Then it says
WARNING! Linpack binary stopped unexpectedly. This could be a result of missing executables, unstable system, or a software bug. If you have UAC enabled, please try re-running the program with administrative privileges.

Test executed for 0.33 seconds.
Click to expand...

1124 MB is the largest value it will let me run with.

Thanks Diggsy, I may well give them a try. That's for 3.6GHz right?
Bit worried about the PWM temp so not sure I'll go that high.
 
Quick update.
I ran IBT 10 times using 4 threads with the memory set to 1124 MB and it passed.
Max PWM temp displayed was 67ºC, max CPU temp was 54ºC (from uGuru app).
This was at 3.0GHz, can I consider this stable or does the fact IBT wont run with a memory setting of over 1124 mean that it's not stable.

I upped the CPU VCore to ~1.3v and set the RAM to 1.8v (uGuru was reporting it set to 1.95v when using Auto).
CPU-z reports the RAM timings as correct.
I'm now running 4 x 1GB stick by the way.

EDIT: Not sure if this test was valid, but I ran 2 instances of IBT, each using 1124 MB RAM and running 2 threads (10 times each). I figured this would run twice the memory. However each run in each instance took twice as long and judging by the Performance tab in Task Manager, if you run IBT with just 2 threads it doesn't run 2 cores at 100% but preads it out across all 4.
 
Last edited:
If you are aimming for 3.6GHz, using 9x400 with the memory running at 800MHz would be the most hassle free.

I have 4 sticks of 1GB memory too. In terms of voltages...as already mentioned above, you need to give the NB a little bump, and as for the vcore, 3.6GHz would typically need around 1.46~1.5v to stable (depending on the chip).
 
Last edited:
Hmm, that's not too hot really. Can you run Prime without restarts? I always find IBT stresses the CPU a LOT more than Prime.
 
projects14a said:
What temp is too high for the PWM? Im trying to get mine running at 3.6Ghz and have not had much luck so far.
Click to expand...

I've just looked through my old posts and when Priming I was seeing 75c or so from the PWMs, and that is with the bolt mod and a 60mm fan hovering over them and it was very stable.

IIRC the absolute max temp is something like 110c, but I would not want to go anywhere close to that!! I would say anything below around 85c would be ok, obviously the lower the better.

Are you running 2 sticks of RAM or 4?
 
Thats cool then, so far they havent gone above 70c.

Im running 2 sticks and running the setting you've stated above. For some reason Prime fails after about 15mins (running blend test). Core temps are around 30c to TjMax, which I believe is safe?

My processor has a VID of 1.2625v and is revision G0 and stepping B. I dont think its far off running at that speed, but Im not sure what changes to make next.

edit: I can get some G Skill Ram which I originally bought for this set up, but thought the OCZ would be better due to its better specs.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom