try 400xthe max multplier u can do and run the ram at 1:1 at the tightest timing u can do and i'll match it and we can see the cache different.

found some reference



Source

thats why i ask 8tvt to run the as i suggest.

ya man.

this 65nm SOI gonna make my power consumption rise as high as the temp.

Overview

There are 4 tests needed to be done before running Linpack Benchmark that usually guarantees 99% success of passing 50 runs. The 4 tests will approximately take a total of 3 hours time. Linpack Benchmark is regarded as one of the the most stressing and demanding benchmark/stability test in the OverClocking Community especially in XtremeSystems

Setting up

All 4 Tests must be run in Windows Diagnostic Mode with Virtual Memory disabled. It is meant to fully test your hardware stability. Screensaver is to be disable and set the Power Options to Performance. Also set the 'Turn off display' to Never. All the test is to be carried out with sufficient cooling for your CPU, Memory, NorthBridge and so on. Besides that, you also need good air flow environment and a stable Power Supply.

Test Tests

QUOTE
Test 1
Vcore: Max
VTT: Med
GTL: Light
NorthBridge: Low
RAM: Low

Start SP2004 Orthos
Set Test option to 'Blend - stress CPU and RAM'
Still under Test option, Click 'Custom...'
Under Custom Test, change 'Min FFT size' to 8, 'Max FFT size' to 8
Under Priority Option, change to 10
Run SP2004 Orthos

Make sure SP2004 Orthos is kept running and passed approximately 35 minutes.
In order to achieve that, you need to run two rounds of 15 mins SP2004 Orthos. After that run Test 2 to make sure it can start the third iteration of 4096 and the quit.

Note: You need to keep increasing your vcore until you did not see see any difference or to pass this test.

Test 2
FSB: Max
RAM: Med
NorthBridge: Med
Vcore: Low
GTL: Low

Start SP2004 Orthos
Set Test option to 'Blend - stress CPU and RAM'
Still under Test option, Click 'Custom...'
Under Custom Test, change 'Min FFT size' to 4096, 'Max FFT size' to 4096
Under Priority Option, change to 10
Run SP2004 Orthos

Same as above, you need to pass approximately 33 minutes with two rounds of 15 minutes. After that run Test 2 to make sure it can start the third iteration of 4096 and the quit.

Note: You need to keep increasing your Northbridge and FSB Termination voltages until until you did not see see any difference or to pass this test. FSB Termination voltage is to be maintain 1 level above NorthBridge voltage, especially for Intel P45 Chipset Mainboard.

Test 3
GTL: Max
FSB: Max
PLL: Max
Others: Med

Start SP2004 Orthos
Set Test option to 'Blend - stress CPU and RAM'
Still under Test option, Click 'Custom...'
Under Custom Test, change 'Time to run each FFT size (in minutes)' to 1
Under Priority Option, change to 10
Run SP2004 Orthos

Make sure to pass all the SP2004 Orthos FFT tests which is from 8 to 4096. After that run Test 2 to make sure it can start the thrid iteration of 4096 and the quit.

Note: You need to keep increasing your GLT and FSB Termination voltage until you did not see see any difference or to pass this test.

Test 4
NorthBridge: Super Max
Ram: Super Max
Others: Low

Run Memtest for Windows
If you are running 4Gb of Memory, run 2 instances of memtest each at 1700. Make sure you passed 100%.

Note: You need to keep increasing your Vdimm voltage and tweak your Northbridge settings until you did not see see any difference or to pass this test. Through Northbridge tweaking, you will know when there is affect to vdimm increase. For example, you keep on getting an error at a certain level.


Attention

All the tests will take up and test 1.5gb of ram. Therefore, sometimes error occured may be just allocation issue. So users need to beware that sometimes SP2004 Orthos detects more free available ram but Windows does not have enough. The thoughest of all is Test 2. you may passed Linpack Benchmark, Memtest 86+, Memtest HCI and Prime95 but you may not pass Test 2.

- Credits to cstkl -

try that and it'll be rock stable.

1 thing bout the test is that if u think the voltage is too high but it help to stabilise the oc then let it be and dont worry bout it but make sure u got proper cooling.

do as the test say and it'll be fine.

up the NB and FSB termination voltage.

dont forget its a quad and the mobo ur using is not the best overclockable mobo in the world.

---

Added on June 5, 2009, 4:40 pmlast time i stabilise an E8400 at 4gh@500x8 it was pretty easy but a 400x8 Q6600 is a different story.

took me a while to make it stable.

This post has been edited by hilmiangah: Jun 5 2009, 04:40 PM

400x8 need 1.36v.

400x9 need a whopping 1.45v to be benchmark stable.

---

Added on June 5, 2009, 4:48 pmstabilising quad need skills man.

ask cstkl1 for help.

This post has been edited by hilmiangah: Jun 5 2009, 04:48 PM

no regret man.

at last i can oc the legendary G0 quad.

this is a must oc proc for ocer imo.

is it really stable??

depends on individual.

some might prefer 8 hours prime while others let it run for 1 hour and satisfied with it.

btw on LinX,try to use all of the ram available and then run the test.772mb just doesnt cut it.

run ur windows in diagnostic mode and u can use up to 97% of ur available ram to stress the cpu.