GA-EP45T-DS3R DDR3-1600 So far so good

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Yeah,
I have skimmed the article by Lsdmeasap and had it in the back of my mind while trying the overclock - but ended up fixated on the advanced timings - since through testing it seemed the system was fairly happy with the primary timings.
I am going to try some voltage manipulation based on Lsdmeasap's great thread.

I will let you know. Thanks for reminding me.

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

OK,
I am confuzzled here, since Vtt is tied directly to MCH ref, whenever you reset Vtt back to original - MCH ref follows.
The prevailing wisdom is to keep Vtt .04 ~ .06 below VCore.

What am I missing here?

Disregard ^^^^^^^

I had one of those "DOH!" moments.

Now for some testing......

[gampamu]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

The MCH/CPU reference settings are not really voltage settings at all. They are settings relative to the VTT and best expressed as a percentage of the VTT rather than as voltage figures. Kinda wish Gigabyte used percent of VTT for MCH/CPU reference settings instead of voltages.

I am not sure when and where the assertion that one must keep the VTT 04 ~ .06 below the Vcore originated but it does not apply to the modern CPU's and chipsets.

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Info about Vtt with regards to CPU V.I.D. (CPU Vcore) came from ProfJim in this thread here:

http://forums.tweaktown.com/f69/ga-e...tml#post305212

Looks like in my original 425 setup I incorrectly stated the RAM at 1.8V - the correct voltage was 1.84.

Still testing......

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Looks like - so far - it will be stable @430 with RAM@ 1.86V. And here I thought it was a timing issue - live and learn I guess.
Stable @425 with RAM@ 1.84V

Now I am going to see if I can drop VCore down a hair using Lsdmeasap's formula.....

[gampamu]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

I believe what ProfJim said in the post was meant to help you with the problems you were experiencing at the time, not intended to be rules for you to follow for all overclocking situations.

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Ah, I gotcha.

So far it looks like MCH ref @61% actual is working.

[gampamu]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

That is very interesting. Out of a gut feeling there may be some difference between the DDR2 and DDR3 P45 boards, I deliberately avoided posting the MCH ref setting that works on my system for that reason. Looks like I did the right thing since the optimal MCH ref setting for my system is 67% ([email protected]). If the optimal setting for your DDR3 board is 61% ([email protected]?), recommending the 67% setting would not have been helpful at all.

[profJim]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

The general VTT (aka Cpu Termination or FSB Voltage) recommendation is that it should be at least 0.05 - 0.10 volts lower than Cpu Vcore.

A google search using:
vtt fsb voltage
will provide a lot of information, some of it will be contradictory.
Some of the reading will be very deep!

Understanding GTL Reference Voltage is one of the better articles.
Don't get hung up on the details, just grasp the overall picture.
Page 7 shows an example where the proper setting increased Prime95 stability from 47 minutes to nearly 13 hours before the test was stopped.

Do share your results, making sure to follow Lsdme's recommendations for properly setting MCH Reference as it relates to VTT.

I don't know if the general VTT and MCH Ref settings differ when using DDR3 memory.

Here's a table that should help when adjusting MCH Ref:

% of 1.20 values [before auto-adjustment]
55% = 0.660 || 60% = 0.720 || 65% = 0.780 || 70% = 0.840
56% = 0.672 || 61% = 0.732 || 66% = 0.792 || 71% = 0.852
57% = 0.684 || 62% = 0.744 || 67% = 0.804 || 72% = 0.864
58% = 0.696 || 63% = 0.756 || 68% = 0.816 || 73% = 0.876
59% = 0.708 || 64% = 0.768 || 69% = 0.828 || 74% = 0.888

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

It looks like 62% may be it, although it seems that a CPU V.I.D. of 1.31875 is the absolute lowest I can go - no matter which way I go with MCH ref.

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Thanks for the additional info ProfJim, this has been an extremely productive thread.
It has definitely helped me, and hopefully it will help others as well.

[hank123]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

has anyone been able to go over 1600Mhz with out using a 3:5 divider?

[profJim]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Using the 4.00 memory multiplier:
cpu = 400MHz x 4.0 = 1600MHz memory speeds
cpu = 410MHz x 4.0 = 1640MHz memory speeds
cpu = 420MHz x 4.0 = 1680MHz memory speeds
cpu = 430MHz x 4.0 = 1720MHz memory speeds
cpu = 440MHz x 4.0 = 1760MHz memory speeds
cpu = 450MHz x 4.0 = 1800MHz memory speeds

Using the 3.20B multiplier, you can reach similar speeds if your cpu host frequency is 25% higher, for example:
cpu = 500MHz x 3.2 = 1600MHz memory speeds

[ghost707]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

Hank,
let us know if you get your RAM past 1400 - and give us all the details.
Creamsoda and I are in the same boat; having to run slower frequencies but compensating with tighter timings - seems to even out in the wash.

[hank123]

Re: GA-EP45T-DS3R DDR3-1600 So far so good

oh ya dude 1600 is cake. anything over is not. GUYS DO NOT USE THE 3:5 DIVIDER. It is bugged. Run everest and see for your self.
The E8500







Now the Q6600



This is the best I have done.



875Mhz BUT ITS ON A 3:5 and the bandwidth sucks ass bad!