A 70 percent overclock - 2.4 to 4!

I am very impressed with this budget overclock. The Q6600 is an amazing chip and the X38-DQ6 really takes advantage of it. Although the system was Prime95 stable on all four cores and temperatures stayed well under 35 celcius, 3DMark06 CPU tests were really pushing this system hard. Frame stuttering was evident on the SM 2 and SM 3 benchmarks. Because temperatures stayed so low I really think the system could run at higher speeds with the ghetto .. err budget cooling set-up. The memory which is only designed for 800 runs was pushing 890.

Here are the steps taken to hit 4 GHz

1. Acrtic Silver 5 - the cpu was covered with as thin a layer as possible and a small drop was place in the middle. The water block was pressed on top and moved around to insure contact with no bubles or dry spots. There were definately runs where specific cores ran hotter indicating some installations did not have adequate cpu to block adhesion. Always run new installs at conservative levels and check temperatures for each core seperately at idle and at load. The CPU water block was reinstalled and temperatures were now low and consistent.

2. Cooling - to insure proper air flow the side cover was left off and fans were placed above the CPU, NB, and memory

3. Video card - since the target was high CPU scores not SM 2 and SM 3 scores, I left the card at stock speeds and stock volts.

4. CIA 2 was turned off in the bios and the PCIe bus was locked to 100.


5. Memory was manually loosened to 7-7-7-22, and voltage was raised to 2.1, the rating from Corsair.













6. CPU speed was set to 350, and voltage was set to the Intel listed 1.5v. I ran Prime95 to make sure the temperature range for the cores at idle and load were within range (7 and 22.)


7. The system was then turned off and the wiper fluid and ice was added to the immersion bin. A small fountain pump moved the cold water from the bin into a simple manifold attached to the radiator with wire ties. Adhesive foam stripping was used to minimize leakage.





8. After covering the ice with salt and immersion fluid was stirred and allowed to chill down to -4 celcius.

9. The system was started up again and in the bios the FSB speed was set to 400 (3.6 GHz with the Q6600 9 multiplier.) Memory is locked to 2.00 which is a very comfortable 800 for the PC-6400C4 chips.

10. Once the system boots in XP I used Gigabyte's EasyTune to monitor temperatures and slowly bump up the FSB and run 3DMark06. This lasted until I hit 415.

11. In the Bios I pushed the NB voltage up, and gave the FSB a mild increase involtage as well. The key here is to push as few volts in as possible to keep down temperatures, and yet bump volts up as needed to maintain stability. The secodn you bump up votage and see no stability increase back off and try something else. Always be sensitive to maximum voltage for your various parts and monitor all temperatures.

12. Back in XP I again bumped up the FSB in Easytune and ran Prime95 to check temperatures. The machine never went over 40 degrees celcius with this setup but Prime95 was not stable over 465. So I jumped in and ran 3DMark06 again.

13. 3DMark06 had no problem with SM2 and SM# tests but the CPU benchmark would fail until I settled at 445. This FSB speed translated to a 4005 CPU speed giving me the 6181 score found here.

http://service.futuremark.com/compare?3dm06=4778765

I am sure the with better memory the cooling setup could have pushed me up to at least 475 FSB. I am very impressed with Intel, Gigabyte, and Coolermaster's contributions to this project, and when I upgrade my memory it will probably be some Corsair PC-10000.

The next task will be closing the box up and finding a speed which allows me to run 24/7 stable at a reasonable temperature - without the ghetto .. err, budget immersion system!

The video will be up soon!