GA-EP45-UD3P - Won't boot with 450FSB

[D-an-W]

I am still tRD 9, but everything else is auto...will update template tomorrow.

[devoh]

D-and-W I'm curious... In CPU-Z under the mainboard tab.. what revision does it say for your chipset? Mine says A3 for my version 1.1 GA-EP45-UD3P.

-devoh

[D-an-W]

A3 for me too.

[devoh]

D-and-W I am having issues with stabalizing a 8 x 515FSB = 4.12Ghz..
Can you post a stable template for something close to 4.12? I know my memory and MCH are correct and stable.

-devoh

[D-an-W]

At work at the moment but I posted my 510 template a few days ago which passed everything I threw at it.

Will post my 530 template this evening...

[devoh]

Cool, I'll go back and find it, thanks.

I bumped my CPU skew to 350 and am having better luck now. So I have a CPU skew of 350 and a MCH skew of 200.. for my 515FSB x 8 = 4.12.. I should post my template of what I am working with.. odd how my cpu seems to be much more tempermental that your E8400. One would assume that a E8500 E0 stepping would be more stable than a E8400 E0 stepping, but obviously that is not necessarly the case.

-devoh

[jonathankru]

Hi All. Looks like you guys have been having allot of success with getting higher FSB. I have managed to boot my Q9650 @ 525 FSB . But i have not played with the MCH and CPU skews. Could someone explain what these change? I hate changing things without understanding what it does. Does it affect performance?

Thanks in advance.

[D-an-W]

I haven't used skew yet Jonathan, just going to grab a shower then will post my latest template...

[devoh]

Skews adjust the fine timing of the clocks. My understanding is that they delay the clock signals by a certain amount. So it doesn't impact performance, as the clock is not slowed down it is just delayed. As you ramp up in clock speed some CPU's and MB's (MCH's) need a little extra fine tuning to be really happy (The clocks get out of sync or jittery) . The CPU Skew helps with high CPU overclocks. The MCH skew with high MCH overclocks. I got a 1.26 MCH voltage to work with a 200 MCH skew and a 515FSB. They are very tricky to use, and may not be needed by everyone.

I have a very unorthodox method that I used to find what my system wanted. I set the FSB to what my goal was (but underclock the CPU).. The machine would not POST.. GOOD, yes good... then increment the MCH voltage until it does POST. Then back the voltage back one so that it will not POST. Now take the MCH skew and increase it (it goes by 50's) keep going until it POSTS, write down that value and keep going until it fails to POST. Your optimum MCH skew is in the middle. Like on my I got a post at 150, 200 and 250 I believe. 200 was the best for me. I then left the MCH skew set to 200.. then start backing the MCH voltage back down until it fails to POST again. Now try adjusting the MCH Skew up and down a few to see if it POSTS again. If it does that use that value and repeat until it doesn't make a difference, that is a keeper. Then go about your normal testing methods for stability.

Then tackle your CPU with a similar method after you are sure your FSB is rock solid.

I have found that 200-250 MCH skew for 500-550FSB is probably right... and 300-350 CPU skew for around 4.1-4.2Ghz.. Naturally just because it works for me, does not mean it will work for you. Good Luck.

-devoh

[jonathankru]

Originally Posted by devoh

Skews adjust the fine timing of the clocks. My understanding is that they delay the clock signals by a certain amount. So it doesn't impact performance, as the clock is not slowed down it is just delayed. As you ramp up in clock speed some CPU's and MB's (MCH's) need a little extra fine tuning to be really happy (The clocks get out of sync or jittery) . The CPU Skew helps with high CPU overclocks. The MCH skew with high MCH overclocks. I got a 1.26 MCH voltage to work with a 200 MCH skew and a 515FSB. They are very tricky to use, and may not be needed by everyone.

I have a very unorthodox method that I used to find what my system wanted. I set the FSB to what my goal was (but underclock the CPU).. The machine would not POST.. GOOD, yes good... then increment the MCH voltage until it does POST. Then back the voltage back one so that it will not POST. Now take the MCH skew and increase it (it goes by 50's) keep going until it POSTS, write down that value and keep going until it fails to POST. Your optimum MCH skew is in the middle. Like on my I got a post at 150, 200 and 250 I believe. 200 was the best for me. I then left the MCH skew set to 200.. then start backing the MCH voltage back down until it fails to POST again. Now try adjusting the MCH Skew up and down a few to see if it POSTS again. If it does that use that value and repeat until it doesn't make a difference, that is a keeper. Then go about your normal testing methods for stability.

Then tackle your CPU with a similar method after you are sure your FSB is rock solid.

I have found that 200-250 MCH skew for 500-550FSB is probably right... and 300-350 CPU skew for around 4.1-4.2.. Naturally just because it works for me, does not mean it will work for you. Good Luck.

-devoh

Thanks. I will give your method a try. I normally do something similar in setting the GTL`s and VTT.

The reason i ask is because i think these skews will help stabilize my cpu @ 4.5Ghz. I have it 3d stable @ 4.65Ghz but i cant run blend without it causing problems at 4.5Ghz. And i have tried every mch voltage and mch reference under the sun.

Maybe this will be the setting that helps me get that last bit of stability.

Thanks Again.