How to set up GTL Ref Values for 45nm & 65nm

[Deathspine]

Trying to fine tune my voltages to reduce the NB temperatures. NB Voltage at 1.58V, NB GTL REF Ratio currenlty at +40 determined by trial and error. Currently reading this forum to learn more about GTL settings. If I got this right GTL are based on CPU VTT. My CPU VTT is 1.31

1.31 x 0.0667 = 0.87377
1.31 x 0.0635 = 0.083185
0.087377 - o.o83185 = 004192
NB GTL = +.04mv

Is that correct?

[CryptiK]

If you want 0.87v for the NB Vref and have a fixed multi of 0.635x and your Vtt is 1.31v, yes.

[Deathspine]

If you want 0.87v for the NB Vref and have a fixed multi of 0.635x and your Vtt is 1.31v, yes.

Thank you very much for helping me out here. I am not sure of your answer "If you want 0.87 for the NB Vref" I thought I was following the "formula" for the correct GTL for the NB based on CPU Vtt. Do I not want 0.87v for the NB Vref?

I have the NB V at 1.6 to get this motherboard stable. I have tried NB GTL at +40, -12 both passed OCCT as stable. I am currently testing +0.04.

[CryptiK]

I meant as you have a fixed NB GTL Ref of 0.635x, by adding 40mv you are simulating having a NB GTL Ref of 0.667x.

Try anything from 0.545x to 0.70x, so Vrefs from 0.71v to ~0.9v.

Eg:
0.545x 1.31 = 0.714v
0.635 x 1.31 = 0.832v
0.714v - 0.832v = -118mv

so to simulate choosing 0.545x you'd choose the closest thing to -118mv

I actually found that my P5Q-Deluxe would prefer really low GTL Refs, other boards prefer higher ones, it just depends.

[moogle]

On the Asus Maximus Formula II for the CPU GTL it has 4 (0-3) ref's to set. The first one uses the formula with 0.635 and then it alternates with 0.667 for the rest of the time. So it looks like this.


0.635
0.667
0.635
0.667


Thats just the formulas it states. So I'm confused as to what to set. It's +40mV for the first one and 3rd one but what about the other two?

Cheers for any help.

[mikeyakame]

generally a little higher values for 0/2 work better. 1/3 don't benefit from as much offset since they are already using a base of 0.667

[moogle]

generally a little higher values for 0/2 work better. 1/3 don't benefit from as much offset since they are already using a base of 0.667

Ahaa now I get the idea. If there was a thanks button I'd use it

Thanks!

[CryptiK]

On the Asus Maximus Formula II for the CPU GTL it has 4 (0-3) ref's to set. The first one uses the formula with 0.635 and then it alternates with 0.667 for the rest of the time. So it looks like this.


0.635
0.667
0.635
0.667


Thats just the formulas it states. So I'm confused as to what to set. It's +40mV for the first one and 3rd one but what about the other two?

Cheers for any help.

For Vtt around 1.2v, I use

+30
-10
+30
-10

for the 4 values on my M2F.

You could also use

+40
auto
+40
auto

or

+50
+10
+50
+10

[moogle]

For Vtt around 1.2v, I use

+30
-10
+30
-10

for the 4 values on my M2F.

You could also use

+40
auto
+40
auto

or

+50
+10
+50
+10

Yeah I'm using 40-auto combo. Although I might try 30 -10. I'm on 1.25vtt

[Deathspine]

I meant as you have a fixed NB GTL Ref of 0.635x, by adding 40mv you are simulating having a NB GTL Ref of 0.667x.

Try anything from 0.545x to 0.70x, so Vrefs from 0.71v to ~0.9v.

Eg:
0.545x 1.31 = 0.714v
0.635 x 1.31 = 0.832v
0.714v - 0.832v = -118mv

so to simulate choosing 0.545x you'd choose the closest thing to -118mv

I actually found that my P5Q-Deluxe would prefer really low GTL Refs, other boards prefer higher ones, it just depends.

Thanks again, I will do some testing with you calculations.

[vox]

On the Asus Maximus Formula II for the CPU GTL it has 4 (0-3) ref's to set. The first one uses the formula with 0.635 and then it alternates with 0.667 for the rest of the time. So it looks like this.


0.635
0.667
0.635
0.667


Thats just the formulas it states. So I'm confused as to what to set. It's +40mV for the first one and 3rd one but what about the other two?

Cheers for any help.

generally a little higher values for 0/2 work better. 1/3 don't benefit from as much offset since they are already using a base of 0.667

I´m getting all bangup with the maths!

If the cpu is as Q9550 (45m) and a P5q3 that as cpu gtl 0/2 and 1/3 with:

Vcore: 1.40
cpu gtl 0/2: auto
cpu gtl 1/3: auto
Vpll: 1.60
vfsb: 1.46
vdram: 2.10
NB gtl: auto
Vnb: 1.38
Vsb: 1.10
VPcie: 1.50

i cant get write the multi. Y= (vtt*0.670)-(vtt*0.630)

then Y+0.630= Z value that i will use put on the cpu gtl =0/2 AND cpu gtl= 1/3????

OR Y+0.630= cpu gtl 0/2 AND Y+0.670=cpu gtl 1/3 ??????

can someone PLEASE run the maths for me with steps because my values are all wrong....?!

Tks!
vox

[truehighroller]

Well, it is a little list of hints.

1) I read mikeyakame's posts.

2) Usually, to get things stable, you HAVE to set the VCore in that voltage range. And that's naturally higher than VCore. In my case 1.32V is enough to get my CPU stable at FSB 460. And anything below VTT 1.34V is unstable. Wellp...

3) I have read a few reports from people who killed their new 45 nm CPUs with FSBT values that they used previously on ther 65 nm chips. Generally beyond 1.45V.

Proof. Well,this is a bit silly. You were asking me for proof elsewhere before where one can hardly provide it.
We are talking about a few experiences, and my 17 years of making my own PCs give me a feeling that i am trusting. And frankly I have enver killed a CPU with overclocking. Must have been a dozen or more by now since 1996. You may or may not trust my judgement...up to you.

I still remember when I was unsoldering the 20 MHz oscillator on an Acer 286 mainboard, replacing it with a 24 MHz type. My first PC overclock....

Right Right. I have mine set to vFSB 1.22 and my gtl reference settings set to .63 and I have my chip super stable @ 3.85GHZ 1.258~v. I was just wondering what lead you two to believe that if you have the vFSB set higher then the CPU Voltage that you run less of a chance of blowing out your CPU that's all. Again not argueing sorry, don't take what I say the wrong way please.

[truehighroller]

I´m getting all bangup with the maths!

If the cpu is as Q9550 (45m) and a P5q3 that as cpu gtl 0/2 and 1/3 with:

Vcore: 1.40
cpu gtl 0/2: auto
cpu gtl 1/3: auto
Vpll: 1.60
vfsb: 1.46
vdram: 2.10
NB gtl: auto
Vnb: 1.38
Vsb: 1.10
VPcie: 1.50

i cant get write the multi. Y= (vtt*0.670)-(vtt*0.630)

then Y+0.630= Z value that i will use put on the cpu gtl =0/2 AND cpu gtl= 1/3????

OR Y+0.630= cpu gtl 0/2 AND Y+0.670=cpu gtl 1/3 ??????

can someone PLEASE run the maths for me with steps because my values are all wrong....?!

Tks!
vox

@VOX


Try my settings mentioned on yours and see if you are stable,


Vcore: 1.40 leave it and mess with the ones below and see if you are stable then try to back this down,

cpu gtl 0/2: auto set these to .63
cpu gtl 1/3: auto .... .63
Vpll: 1.60 set this to 1.56
vfsb: 1.46 try 1.22 like mine is
vdram: 2.10 Overvolting right now drop it to 2.02v and see if it is ok.
NB gtl: auto
Vnb: 1.38 Set this to 1.30v
Vsb: 1.10 Set this to 1.20v as it is overvolting your south bridge set like you have it right now "bad".

VPcie: 1.50 same as the sb overvolting, set this to 1.6



As far as the math goes you are aiming for supposed between, .80 - 1.05 with the equation being your gtl reference say .63 like mine is times the vFSB 1.22 like mine is so .63 x 1.22 = .7686. That is what mine is set to and I'm more stable at lower voltages then alot of people... Not every chip is the same yada yada.. exc..

[vox]

@VOX


Try my settings mentioned on yours and see if you are stable,


Vcore: 1.40 leave it and mess with the ones below and see if you are stable then try to back this down,

cpu gtl 0/2: auto set these to .63
cpu gtl 1/3: auto .... .63
Vpll: 1.60 set this to 1.56
vfsb: 1.46 try 1.22 like mine is
vdram: 2.10 Overvolting right now drop it to 2.02v and see if it is ok.
NB gtl: auto
Vnb: 1.38 Set this to 1.30v
Vsb: 1.10 Set this to 1.20v as it is overvolting your south bridge set like you have it right now "bad".

VPcie: 1.50 same as the sb overvolting, set this to 1.6



As far as the math goes you are aiming for supposed between, .80 - 1.05 with the equation being your gtl reference say .63 like mine is times the vFSB 1.22 like mine is so .63 x 1.22 = .7686. That is what mine is set to and I'm more stable at lower voltages then alot of people... Not every chip is the same yada yada.. exc..

ok True, but i need to learn how to calculate this the correct way!

Can you please post the exact Mathmatical steps to get to this values?

Help!

[truehighroller]

ok True, but i need to learn how to calculate this the correct way!

Can you please post the exact Mathmatical steps to get to this values?

Help!

GTL reference x vFSB = Number you want to have between .80 - 1.05 according to Intel,,,,


That is it, so like I said with me it works out to be,

.63 x 1.22 = .7686 and mine is stable 100% @ 3.85GHZ 1.258v~.

[vox]

GTL reference x vFSB = Number you want to have between .80 - 1.05 according to Intel,,,,


That is it, so like I said with me it works out to be,

.63 x 1.22 = .7686 and mine is stable 100% @ 3.85GHZ 1.258v~.

ok but .7686 isnt the number you are going to put on the gtls.

what it is it?

is .7686 + GTL 1/3 (0.667) ? or .7686 - GTL 0/2 (0.650)?

( vtt*gtl 0/2 ) - ( vtt*gtl 1/3) = Y them (Y + GTL 1/3) and this is number to put on the GTLs?

or

( vtt*gtl 1/3) - ( vtt*gtl 0/2 ) = Y them ( Y + GTL 0/2 ) and this is number to put on the GTLs?

How do you get to the 0.630 GTL Refs you are telling me to put????

[truehighroller]

You are reading to far into it... The formula for us because you have a p5q as well.. is , gtl ref x vFSB and that equals the number that matters to us,,,,, which according to Intel,,,, should equal between .8 - 1.05 but,,,,, mine is 100% stable and people that have tried my settings with similar quads have had success as well with my settings.... which as we both know and I have posted it three times now,,,, = .7686 but, it is working and I have a low voltage core lower then not everyone but, the majority of people with Q9450s and alot of other .45nm quads for that matter... Just give it a go.......

[carepolice]

ok but .7686 isnt the number you are going to put on the gtls.

what it is it?

is .7686 + GTL 1/3 (0.667) ? or .7686 - GTL 0/2 (0.650)?

( vtt*gtl 0/2 ) - ( vtt*gtl 1/3) = Y them (Y + GTL 1/3) and this is number to put on the GTLs?

or

( vtt*gtl 1/3) - ( vtt*gtl 0/2 ) = Y them ( Y + GTL 0/2 ) and this is number to put on the GTLs?

How do you get to the 0.630 GTL Refs you are telling me to put????

Yes you dont put the exact value for GTL's, you need to use Multiper (example, like truhighroller his 0.630x is multiper ! )

You need figure out what multiper you need for e.g. 0.8V

I give fast lesson!

Lets think that your looking for 0.8V gtl's now, and 1st thing that you need to know what VTT/FSB voltage your using!

Lets say that your using 1.24V FSB voltage/VTT

2nd you need to do some very easy math

You are looking for 0.8V GTL for both 0/2 and 1/3 setting

First you try 0.630x X 1.24 (your VTT) = 0.781v

Now you got 0.781 result, But that is not the value that your looking for

Lets put few notch more to our multiper (0.630x, it grows by 0.005x every notch)

You set 0.640x This time to both GTL's, and calculate wich value it gives this time!

0.640 X 1.24v = 0.793 ! We are close now

0.645 X 1.24 = 0.7998 !

0.650 X 1.24 = 0.806 !

This is how you use GTL's, it can be hard to find exact value but all formulas that i have read says that 0.8-1.1 are usually ok?

And now you ended to 0.806! You boot to windows and test does it make any difference

Carepolice man whit too much free time!

[truehighroller]

Thank you, . My point is to though that mine is fine where it is @ vFSB 1.22 and both of my GTLs set to .63 and I have a nice clock so start a little under the .80 starting line and see if that works for you as well first if not, which is possible it might not be stable for you where have mine. Then bump it up a little more like carepolice said until you hit a stable spot..

[vox]

truehighroller

Carepolice

and All the Rest

At last i could understand how to make the math THANKs a lot ppl!!!

Now the next gtl the NB GTL Ref can i use the same formula in the same way?

TKS!

[truehighroller]

You wanna make sure that the ouput number the one that matters that I mentioned, for the nb is not the same as the output number for the vFSB. In other words if they are the same you will have isses with the wave forms bumping heads so to say and you will not be stable. I leave mine on auto and have no issues you might want to do the same thing but, it is up to you. Your welcome as well.

[vox]

You wanna make sure that the ouput number the one that matters that I mentioned, for the nb is not the same as the output number for the vFSB. In other words if they are the same you will have isses with the wave forms bumping heads so to say and you will not be stable. I leave mine on auto and have no issues you might want to do the same thing but, it is up to you. Your welcome as well.

ok True dont get the same values on the mv for the ref cpu and ref nb.

but how do i calculate it?

[truehighroller]

Same way.

[Infa]

This thread is really interesting! like it alot! thanks for all the helpful and people with knowledge who have the patiens to help us "idiots"

allways wanted to learn about GTL, its fun to do this trial and error, and still almost know what ur doin...

thanks again.

[truehighroller]

It is a good thread thank you OP.