Overclocking the Q6600

[Sadasius]

Thought I would start this thread since many of us have one and would like to share some results, settings and some general tips into getting the absolute most stable power out of our CPU's.

Here is a Q6600 B3 stepping going over 4ghz watercooled.

http://www.youtube.com/watch?v=NNm9O_T0Ifw

[Kougar]

I still would like to write up an overclocking guide, but the biggest problem is I only have a half-dozen Gigabyte motherboards + one buggy Foxconn 975X board! Gigabyte does things completely their own way, which while great for overclocking means that I cannot easy take what I know and convert it over to ASUS users, or NVIDIA chipset boards, etc as they are completely different.

I don't pretend to know NVIDIA chipset overclocking, but I would strongly suspect its a chipset and/or FSB related cause. P35 requires +0.2v MCH voltage to avoid locks and roughly +0.1 or 0.2v FSB to maintain long-term stability. tRD and other chipset strap settings are critical at 400FSB with a Quad.

What did you strap yours to? 266, 333 or 400? This was actually something I did not understand and was trying to on some discussions. My Maximus Extreme has so many functions I still do not yet know what the hell they all do or stand for.

***EDIT*** sorry for the thread jack....will start another thread with this topic.

I said all that because to answer your question in the other thread Gigabyte does not offer manual strap settings unless using the 2:1 memory ratio. Obviously my DDR2 RAM will not run at 1600Mhz, so I leave the straps alone as the board auto-handles them. Current I am sure it is using the 400 strap though. According to Intel the MCH strap affects what memory ratios you can use, but 333/400 are the only straps an overclocker should consider. 400 obviously better as it allows a 6 tRD at ~400FSB

My Q6600 G0 runs at 2.4GHz @ 1.10v, 3.20GHz @ 1.20v, 3.6GHz @ 1.43v, and 4GHz @ ~1.55v. Real Temp reports 49-55c temps at 3.6GHz.

400-450FSB is about typical for 65nm Quads to max out within, and appears to be the same for 45nm Quads as well with the occasional exception. The MCH is under a tremendous strain overclocking a Quad even at 400FSB, so the littlest things such as overly aggressive memory settings, an overclocked PCIe bus and the tRD setting can make the difference between it staying stable or going over the edge.

At 3.6Ghz I use 1.43v, +0.2v MCH, +0.15v FSB, tRD of 6, 1:1 memory ratio. Tightening the RAM timings beyond 4-4-4-12 actually forces a higher tRD of 8, which easily hurts performance. This is partly because 4GB of 800MHz RAM already stresses the memory controller. I leave the PCIe bus manually set to 100, since "AUTO" it tries to change itself.

[Sadasius]

I was wondering if we can have this moved to the overclocker's section. This will be a really good place for it as there will be people looking for this thread as the chip grows more and more popular. Not to mentioon I will be posting some pretty heavy overclocks with this chip as well.

[Methious]

I'm running my Q6600 on a P5KC dual memory board.

400 FSB x 8 multiplier at 3.2 GHZ with a set of Mushkin 12800 DDR3 running at 1600 7-7-6-18. 1.26v on the CPU. 1.9v on the ram.

Had to up the NB voltage to 1.55 and pump the FSB to 400 to get the ram up to 1600. I've had it up to 1754 but the NB was screaming for mercy. And at 435 I seem to hit a FSB wall. Typical of the Q6600.

I think I'm going to try the strap at 400. Been working with a guy over at Mushkin who gave me some tips about running the ram at 1800 but I'm not sure the P5KC is going to handle it. They run it at 1800 and have hit 2ghz on a Maximus board.

[Sadasius]

@ Methious What were your temps at that OC? Do you have a high variance in core temps?

Once my rig is done I am going to shoot for 9x450 4.05ghz(Maximus Extreme with custom water cooling). Already have all the settings down. Found one guy that managed to get 4.5ghz on this with a water cooling setup but he was just able to boot and get a screenie of CPUz before it crashed. He had a vcore of 1.808...That's high!!!

[Kougar]

Bloody lucky he didn't crash more than just the system with that much voltage. 3.2Ghz @ 1.20v verses 4.5Ghz @ 1.8v, yeah, I wonder what I'd run at 24/7 :mrgreen:

To be completely honest, Q6600's naturally should reach 3-3.2GHz with very minimal voltages required. But eeking out anything beyond this will face diminishing returns, and anything over 3.6Ghz will just depend on the individual chip and lots of fine tuning. My Q6600 is a native 1.20v vCore, at anything approaching 4GHz my WC loop simply can't wick away the heat quickly enough. 1.5v and 4Ghz will put my core temps around 70c. And I already mentioned in the other thread, those settings will also perfectly double my power consumption @ the wall!

[Methious]

I'm on water with a triple rad and 3 83CFM fans running an Apogee GT waterblock and Micro res.

Generally I idle around 36C and load out at 48C never seen it higher than 49C and that was on a ten hour Orthos run. I can probably trim voltages a little but I got busy with a pile of hardware.

Got a ThermalRight water block and a nexXxos water block to run against it, the ThermalRight might give the Apogee a run but the NexXxos is acrylic and I don't expect much from it.

Waiting on another pile of hardware due in today or tomorrow so it might be a day or four before I get to it.

[Sadasius]

I am only going to try just to get my CPUz screenie and then back it down again to a more reasonable level. My water loop will hopefully hold it steady when I try and shoot for the 4ghz.

What do you have for a WC loop when you did the 4ghz Kougar?

I have the D-Tek Fuzion block for the CPU
EK waterblocks for the moffets, NB and SB
Tygon 1/2" tubbing all around
2X MCP655B pumps
2X GTX240 rads (gen 2's)
EK res with anti cyclone insert.

Basically it will go from Res--->Pump#1--->CPU--->Moffets--->NB--->SB--->Rad #1--->Pump #2--->Rad#2---> back to Res

[Sadasius]

I'm on water with a triple rad and 3 83CFM fans running an Apogee GT waterblock and Micro res.

Generally I idle around 36C and load out at 48C never seen it higher than 49C and that was on a ten hour Orthos run. I can probably trim voltages a little but I got busy with a pile of hardware.

Got a ThermalRight water block and a nexXxos water block to run against it, the ThermalRight might give the Apogee a run but the NexXxos is acrylic and I don't expect much from it.

Waiting on another pile of hardware due in today or tomorrow so it might be a day or four before I get to it.

What kind of rad are you using? Are you just running a CPU block? What is the size of tubing?

***EDIT**** Sorry for the double post. I get excited and carried away when talking about OC'ing and rig builds.

[Kougar]

Right now my WC loop is very lean, I even removed the quick-disconnect PCI bracket pass through as it really improved the flow rate. During my 4GHz run the loop was the same as in my sig, although I had the fans literally screaming since things were quickly getting hot.

Resovoir -> Pump -> CPU -> triple 120 radiator -> res

Something odd that I noted, is my WC loop is not very flow rate dependant. I can drop the pump speed from max down by half (to the 2-3 setting) and temps will not change. If I drop the pump to the lowest setting temps will only increase ~1-3c or so. Logically I can deduce this means my WC loop performance is dependant on something else besides flow rate, and unless I'm pushing the CPU to 4GHz radiator fan speed doesn't affect temps unless I set the fans to their lowest possible setting.

It probably just means that I don't have enough heat in the loop yet, but a future GPU + GPU block should fix that. ;)

[Sadasius]

Flow rate is great with impingement blocks. The more force you throw at the CPU block the better which is why the first thing off of the pump should be the CPU block and not the rad like some people have it setup thinking the fluid is cooler and better for dropping temps but lack pressure after coming out of the rad. Which is why I have a second pump in between the rads to help keep pressure back up and to keep thinks flowing nicely. I don't think you can pinpoint a single thing in a WC setup. It is more of the combined details of the whole that is really the key. People who have taken shortcuts on everything notice it in their temps compared to those that put some thought and care into their loop. The one component that really makes a difference is the rad(s). The type used and what fans you have with it. Let's take the Thermachill for instance. Great flow and does not require a heavy fan to do the job. But there are better rads out their that just require a higher CFM fan(s) and they depend heavily on fan RPM and flow restriction to cool the fluid. So it would be useless to use a impingement block with such a rad. Not the greatest example but it shows that things have to be matched up to what you want your loop to do for the best performance.