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

[CryptiK]

So to simplifier you take your FSB Termination Voltage (VTT) and multiple it by whatever increments you have for GTLs, right?

IE:

45nm:
VTT x GTL increment = GTL Refs

1.22 x 0.635 = 0.7747

Or have I got that completely wrong?

You are correct if you just have the option to set which multiplier is used, and the resultant GTL+ Ref voltage is Vtt x GTL Increment.

Aye we are all here to help each other. I don't think any other spec datasheet lists the GTLREF voltage specs besides the QX9000 one. I haven't seen it in any other datasheet at least.

Thanks mate yeah I just checked through and only found it mentioned there. I actually run my e8400 at 0.667x as I find it is more stable. I find it strange they don't mention it elsewhere.

Thanks for fielding the other questions too.

[Dee]

This looks like a useful thread, although I still don't understand it fully after reading it several times! How exactly should I set my GTL voltages? Details of my current settings are below. If there's any way I can use less vCore or get a slightly higher OC with the same vCore as I'm using, that'd be awesome.

CPU Voltage [1.4250] (real = 1.408 under load)
CPU GTL Reference (0/2) [Auto] (Defaults to 0.630x)
CPU GTL Reference (1/3) [Auto] (Defaults to = 0.670x)
CPU PLL Voltage [1.54]
FSB Termination Voltage [1.20]
NB Voltage [1.26]
NB GTL Reference [Auto] (Defaults to = 0.630x)

Any help on this would be appreciated. Sorry for the n00b post, I never had to deal with these settings when I was using AMD!

[CryptiK]

You just have the multipliers to set so that is easy.

Start by setting both to 0.635x - if this is unstable, increase them one increment at a time until it's stable. I run mine at the equivalent of 0.667x and can run vcore one notch lower, so you may need to set ~0.670x for both of them. It's trial and error to find your cpu's sweet spot.

[mikeyakame]

Generally higher vcore, and higher VTT you can run lower gtlref multipliers depending on clock frequency and multi for cpu. Setting the right GTL reference voltages is all just trial and error. Adjust VTT and adjust the GTLREF multiplier opposite of each other is what I usually do. If I modify VCore without modifying VTT then I adjust the GTLREF multiplier slightly higher, where as If I modify VTT without modifying VCore I will adjust it lower. Same goes for NB voltage, if the bios has NB GTLREF multiplier adjustment, you need to fine tune it when you modify vNB or vTT.

One other thing that I'll mention is. I always keep vTT higher than vCore. This might sound weird to most but it is actually how vTT is designed to work. It acts as a buffer for vCore during overshoot mainly during power on. If it is actually lower than vCore then there is a higher risk you can damage a CPU from too high vCore. It is also basically a limiter for valid BCLK high. IE. I run 1.43v vCore on my Q6600 G0 and 1.42V vTT in my Rampage bios (which is actually 1.47V vTT when loadline is enabled on my particular board).

[CryptiK]

Yes thats true, I use a (comparatively to most) low Vtt of 1.16v so I use higher GTL+ Ref.

[Singh400]

You are correct if you just have the option to set which multiplier is used, and the resultant GTL+ Ref voltage is Vtt x GTL Increment.

Well I've got both my GTLs set tp 0.635 and it's running fine. Don't really see a need to increase it. TBH still trying to get my head round it all.

[mikeyakame]

Well I've got both my GTLs set tp 0.635 and it's running fine. Don't really see a need to increase it. TBH still trying to get my head round it all.

Yeah only change it if you adjust vTT , if its on auto leave GTLREF on auto. You shouldn't adjust GTLREF at all unless you manually set vTT and vNB.

[Dee]

I wish I could get my head around this stuff. I still have no clue.

[mikeyakame]

I wish I could get my head around this stuff. I still have no clue.

Don't worry if you cant get your head all the way around it, it takes time and practice. Lots of reading too! There are still some things I don't have my head completely around because I never bothered to my finish studying my engineering degree! The best way to understand how it plays a role in your system is to try making bios changes gradually and run tests like Linpack, Prime95, Everest etc while watching voltages through Rivatuner, Everest, HWMonitor, etc to get a feel of how different changes affect your setup.

It's one of those things where if you try and understand it without applying it you will end up confused!

I'll try and draw up a 2D or 3D graph in Autocad for you guys to explain it better and create a pdf of it. Ideally i'd like to be able to show the relationship between AGTL+ reference voltage and the associated voltages / limits such as Crossing Voltage, Crossing Point, Min and Max Thresholds, vIL, vIH, Overshoot, Undershoot, FSB SWING, Rising / FAlling edges of BCLK0/BCLK1, BCLK (FSB Base clock), and anything else I can manage to explain visually.

[Dee]

Thing is, I feel these extra settings I know nothing about could be limiting my overclocks. I have possibly the worst E8600 on here. 4.33GHz for 1.40v? That's rubbish compared to pretty much every other E8600 in the database. I would love a real OC pro to set up my BIOS and see what the CPU is actually capable of. Right now, I'm fumbling through it. Having been an AMD person for the last 4 years I guess it's gonna take some time, but man it's quite frustrating at times!

[mikeyakame]

Thing is, I feel these extra settings I know nothing about could be limiting my overclocks. I have possibly the worst E8600 on here. 4.33GHz for 1.40v? That's rubbish compared to pretty much every other E8600 in the database. I would love a real OC pro to set up my BIOS and see what the CPU is actually capable of. Right now, I'm fumbling through it. Having been an AMD person for the last 4 years I guess it's gonna take some time, but man it's quite frustrating at times!

Don't beat yourself up so bad, with time and trial you'll figure out what your particular board and chip likes. You also need to play with Memory Dividers and vPLL they can assist you as well.

[Dee]

OK, I'll keep tweaking. Thanks Mikey.

[fps_dean]

Good post!

I found a real easy way.

VTT = 1.20v (use a nice clean number for ease of math);


To get

VTT * 10 = 12v

Now change v to mv. 12mv = 1%.

So to go from 67% to 63%, you want -12*4=48mv

To go from 67% to 61% you want -12.5*6=72mv

I find this to be easy enough to calculate in my head on the fly using this shortcut method.

Of course, that's assuming GTLVref is 67% as anandtech states rather than .667 by default.

[TouGe]

One other thing that I'll mention is. I always keep vTT higher than vCore. This might sound weird to most but it is actually how vTT is designed to work. It acts as a buffer for vCore during overshoot mainly during power on. If it is actually lower than vCore then there is a higher risk you can damage a CPU from too high vCore. It is also basically a limiter for valid BCLK high. IE. I run 1.43v vCore on my Q6600 G0 and 1.42V vTT in my Rampage bios (which is actually 1.47V vTT when loadline is enabled on my particular board).

Doesn't this then limit the amount of Vcore you can use on 45nm parts? If Intel says that the max vtt is ~1.155V and the Vcore needed for say 3.8GHz is 1.31V (LLC enabled), wouldn't that put out of spec for vtt?

What is more dangerous a high vtt or a high PLL?

I can get 3.8GHz stable (9.5x400) w/ the following settings w/ hardware in sig:
CPU Voltage: 1.31875 (LLC enabled)
CPU PLL: 1.53975V
FSBT: 1.23250V
DRAM: 2.10
NB: 1.23250V

CPU GTL (0): Auto
CPU GTL (1): -45mV
CPU GTL (2): Auto
CPU GTL (3): -45mV

So by what you were saying it would be better for me to increase my vtt to at least 1.32V to help buffer overshoot even though I am stable at a lower vtt?

Also, I don't remember where this was posted in this section, I had read that you want to keep your vtt as close to your NB voltage as possible for stability, was that statement wrong?

[mikeyakame]

Max instanteous vTT is generally the same value as max instaneous vCore +- 5-10%. Intel doesn't recommend over 1.45V continuous vTT on 45Nm parts which is funnily enough the same soft limit for vCore continuous voltage. The buffer effect is why vTT should be soft limited to no more than the maximum safe vCore. A little over won't kill the chip as long as you compensate with a little less vCore

[tekjunkie]

Maybe this would help. I made this for my reference when I had my QX9650 on the S2E.

[Dee]

I wish I could say it helps but I'm just as confused now as I was before I started reading this thread! Nothing makes sense, or I'm too stupid to understand...

[TouGe]

Max instanteous vTT is generally the same value as max instaneous vCore +- 5-10%. Intel doesn't recommend over 1.45V continuous vTT on 45Nm parts which is funnily enough the same soft limit for vCore continuous voltage. The buffer effect is why vTT should be soft limited to no more than the maximum safe vCore. A little over won't kill the chip as long as you compensate with a little less vCore

Okay, where does Intel recommend no more than 1.45V vtt and by soft limited are you referring to what is actually set in the BIOS? Also, when you speak of "max instantaneous" vtt are you telling me that vtt actually fluctuates from the set value during power-on or a switch from low to high load?
Not trying to pick your statement apart but just trying to understand your choice in words.

[tekjunkie]

I wish I could say it helps but I'm just as confused now as I was before I started reading this thread! Nothing makes sense, or I'm too stupid to understand...

I will try to explain. As Everybody has stated, the Intel specs say that quads have a reference GTL of 66.7% or 67%(hence the 0.667 or 0.67) of VTT/vFSB. But for some reason, even if the board gives the GTL voltages as per Intel's reference numbers, the chips are not stable.I don't want to go into "Who's fault is it anyway?" game between the mobo manufacturers or Intel. Remember, this only happens @ high FSBs with the quad.So to make the chips to be stable you think it's the Vcore and increase the Vcore, but still end up with an unstable chip and also unnecessary heat. That's the reason a lot of us with our experience with playing around with GTLs can give you a reference point like saying the 45nm quads seem to be stable with less of a GTL than the refernce of 67% of VTT.

Now, going into the calculation, let's say your rig is semi stable with a VTT of 1.30V. According to the reference from Intel, if you leave the GTLs @ auto, it will give the GTLs exactly that(0.667 times 1.3). But, for 45nms you want your GTLs lower than that. So you determine that you have to give negative values for the GTL so that board will give 0.667 times VTT-xx where -xx is the amount that you actually set in the BIOS of different boards. That's why when you see in my spreadsheet, I had put Auto as zero. Now the question is what does the -xx amount that you would have to specify in the BIOS. It's a simple math calculation. If you want to give a GTL of say 0.657, you are reducing the reference GTL by -0.01. So for a VTT of 1.30V, the equation becomes 1.3x-0.01x1000 = -13mV. The x1000 is because it's in milli volts.For a GTL for 0.647, you go 1.3x-0.02x1000 and so on and so forth.


Sorry about the long post. Hope this helps.

[Big Lar]

Great explanation, heck even I understood that for once Thank You.

Larry

[Dee]

I will try to explain. As Everybody has stated, the Intel specs say that quads have a reference GTL of 66.7% or 67%(hence the 0.667 or 0.67) of VTT/vFSB. But for some reason, even if the board gives the GTL voltages as per Intel's reference numbers, the chips are not stable.I don't want to go into "Who's fault is it anyway?" game between the mobo manufacturers or Intel. Remember, this only happens @ high FSBs with the quad.So to make the chips to be stable you think it's the Vcore and increase the Vcore, but still end up with an unstable chip and also unnecessary heat. That's the reason a lot of us with our experience with playing around with GTLs can give you a reference point like saying the 45nm quads seem to be stable with less of a GTL than the refernce of 67% of VTT.

Now, going into the calculation, let's say your rig is semi stable with a VTT of 1.30V. According to the reference from Intel, if you leave the GTLs @ auto, it will give the GTLs exactly that(0.667 times 1.3). But, for 45nms you want your GTLs lower than that. So you determine that you have to give negative values for the GTL so that board will give 0.667 times VTT-xx where -xx is the amount that you actually set in the BIOS of different boards. That's why when you see in my spreadsheet, I had put Auto as zero. Now the question is what does the -xx amount that you would have to specify in the BIOS. It's a simple math calculation. If you want to give a GTL of say 0.657, you are reducing the reference GTL by -0.01. So for a VTT of 1.30V, the equation becomes 1.3x-0.01x1000 = -13mV. The x1000 is because it's in milli volts.For a GTL for 0.647, you go 1.3x-0.02x1000 and so on and so forth.


Sorry about the long post. Hope this helps.

OK, I think I'm starting to get it. I'm running an E8600, btw. I have two GTL options in my BIOS, (0/2) which defaults to 0.63, and (1/3) which defaults to 0.67. What starting point would you recommend for tweaking these? vTT doesn't usually exceed 1.26v. I also have NB GTL reference which I leave at Auto. I would love to OC my CPU higher but it needs lots of vCore (1.4v for 4.33GHz, which is below average for this chip).

[mikeyakame]

Err let me try again. It's damn confusing to explain but here we go.

There are two particular GTL Reference voltage values you need to be concerned with. These are NB GTLREF and CPU GTLREF. They are two difference reference voltages for two differential clock waveforms occurring seperate of each other. They both have similar purposes but for different clock signals. There are other GTL Reference voltages but these don't have any play on overclocking really, there is only one common factor between all of them, and that is vTT. vTT is the common termination voltage used for CPU, NB, CMOS and PECI GTL+ values which they all need to operate under to work correctly.

NB GTLREF affects the Reference voltage point for the FSB Clock Waveform.
CPU GTLREF affects the Reference voltage point for the CPU Clock Waveform.

Any changes made to vTT must reflect on both NB and CPU GTLREF values. CPU GTLREF isn't effected nearly as much and its minimum Low and High Input voltages are fairly close to GTLREF, they are 100mV or 0.1V off GTLREF. There is a 0.2V swing from low to high and high to low, and provided that the lowest point of the differential waveform doesn't occur within the swing range you are safe.

NB GTLREF on the other hand which negates FSB validity is much less tolerant to differential waveform minimum points. It requires a lower multiplier ratio at higher vTT to maintain the minimum low and high input voltage values.

I need to sit down and try and figure out a simpler way to explain this so you guys can make sense of it. I'll continue this post soon. Need to make a coffee and clear my mind with some C++ code writing.

[Dee]

Thanks Mike. To be honest, I'm almost ready to give up on GTL tuning. I appreciate the explanations, though.

[joger]

I've been reading this thread and I still don't understand this: What do I want my values to be?

-My E8500 C0 needs 1.32v to post at 4.5ghz, but it needs over 1.5v to be 3D-bench-stable at 4400mhz

-9.5x463 is just as stable as 8.5*517, so the instability shouldn't be because of the ram,
because both mem and the mobo's FSB pass SuperPI at 550mhz, and are IntelBurnTest-stable at 533.

Max suicide-shot on 1.40v is 4609mhz @ 9.5 x 485
and on 1.56v 4662.4mhz @ 9.5 x 490.8
so I don't think its about vcore holding it back, something else is wrong or its just a crappy chip ;(

My current 24/7-values are as follows:
7.5x533 ~4005mhz 1.36v (one step down and it won't pass IntelBurnTest), GTLREF on auto
ram on 533mhz 1:1 5-5-5-15 @ 2.21v
NB on 533mhz, 1.47v (one step down and it won't pass IntelBurnTest), GTLREF on auto
My vFSB, of VTT if you will, is 1.26v
and my vPLL is 1.55v

I'd be interested in lovering my vcore 24/7, and achieving more maximum CPU frequency as well as bench stability past 4400mhz. The NB doesn't even get hot to the touch so I guess its fine as it is. I'm running a Asus P5E Deluxe with the Rampage Formula 0408 bios

[Pisklink]

Since a few days I replaced my E8400 with a E8600 (Q822A435, VID 1.21v) due to a lack of OC possibilities. According to the batchnumber this should be a great CPU to OC. However I'm having difficulties to get things stable. I think it could have something to do with GTL settings. In general I understand to principal after reading thetechrepository.com and some other site.

I found that I can boot to 610FSB with everything on auto settings, except for dynamic read / write and spread spectrums. I want 450FSB to get my DDR3 to run at 1:2 divider. With still everything on auto I can boot into Windows and CPU-z reports 1.32v but it is far from stable with orthos. I started fiddling around with VTT, vNB and higher vcore (up to 1.41v in bios) without much succes.

Help would be much appreciated.

OK, I think I'm starting to get it. I'm running an E8600, btw. I have two GTL options in my BIOS, (0/2) which defaults to 0.63, and (1/3) which defaults to 0.67. What starting point would you recommend for tweaking these? vTT doesn't usually exceed 1.26v. I also have NB GTL reference which I leave at Auto. I would love to OC my CPU higher but it needs lots of vCore (1.4v for 4.33GHz, which is below average for this chip).

I got the same options in my bios but how the hell did you find out what your default GTL values are? And how high your VTT really is?