EP45-UD3R Overclocking Question.

[Chike]

Re: EP45-UD3R Overclocking Question.

Good to se you are stable.
Have you ran 5-10 IBT runs?
Memory need a bit of tweaking, tRD(Performance Level in MemSet) is disabled, this has major effect on memory performance. Try setting Static tRead Value 11 for both channels, save, reaboot and try to set it lower, untill the machine won't boot, then set it one step higher.
Also try setting Driving Strength 1066 for both channels.
Do some memory benchmarks before and after to see the differences. Any benchmark will do, I use myself.

I recommend using RealTemp or CoreTemp to monitor temps, both programs record min/max temps.

[ziong]

Re: EP45-UD3R Overclocking Question.

hi. just found something new again.
regarding to the MCH core , currently i was using 1.2V for XP no problem,but when i load Windows 7 Ultimate it needs 1.24V , cause if don't tune it, the OS won't detect the graphic card and the OS will look a little bit Strange。

[profJim]

Re: EP45-UD3R Overclocking Question.

With 2x1GB of memory, MCH Core = 1.24 seems to be high for DDR2-800 memory, but I wouldn't worry about it.
Have you run extended stress tests with both operating systems?

[ziong]

Re: EP45-UD3R Overclocking Question.

With 2x1GB of memory, MCH Core = 1.24 seems to be high for DDR2-800 memory, but I wouldn't worry about it.
Have you run extended stress tests with both operating systems?

yes, both tested by Intel Burn Test, passed.
currently i use another new setting with lower V dimm and lower v Core from the previous.
can IBT trusted by stressing system in few minutes and consider your system is stable?

[Chike]

Re: EP45-UD3R Overclocking Question.

I wouldn't trust IBT alone for stability.
You can try and lower vcore further untill GFlops results drop, then set it back one or two steps up. You can do this with EasyTune but final settings should be made in BIOS abd re-tested. Try one step up PLL and see if that help with lowering vcore too.
Is tRD 10 the lowest you could get?

[ziong]

Re: EP45-UD3R Overclocking Question.

Is tRD 10 the lowest you could get?

will try tRD 9 later, quite surprise that you can do tRD7 @800 (4-4-4-12)
i have google up my ram ,google say it is 90% using D9GMH chipset, but i doesn't open the sink, cause it will void my warranty status.

[profJim]

Re: EP45-UD3R Overclocking Question.

Using tRD = 9 and 2.00D memory multiplier with FSB = 445MHz should be stable.
Using tRD = 8 and 2.00B memory multiplier with FSB = 445MHz should be stable.
tRD = 7 is generally usable with EP4x mobos up to about 420MHz.

IBT or LinX for 10 loops using maximum available memory is the first stress test that I perform.
If the test fails, I don't perform additional stress tests until I've done additional tweaking to fix the instability.
I've had many tests where IBT or LinX were error free, only to have Prime95 or OCCT fail stability testing.
Prime95/OCCT need to be run for at least 8 - 12 hours, as errors can show up with the longer tests, sometimes from 12 - 24 hours.
Click on the Prime95 Tutorial link in my signature for information.

[Chike]

Re: EP45-UD3R Overclocking Question.

will try tRD 9 later, quite surprise that you can do tRD7 @800 (4-4-4-12)

tRD depends on bus frequency, bus/ram frequency ratio and CAS latency. It is common for DDR-800 CL4/DDR-1066 CL 5, to run tRD 7 at 400MHz FSB.
I am quite sure 9 will work for you, or even 8.

I have once found this formula in The Rules of Working with tRD
It is for X48 chipset but seems to be ok most times

For you current FSB/RAM ratio:
445FSB DDR-890, x = 2, N = 1, tCL = 5, tRD = 8
8 - 5/1 > 2/1 ?
3 > 2 ?
answer is yes

I must add that at least one time I have found it to be wrong (I don't remember the settings though)