P45 and increasing PLL Voltage

[GrandMasterJ]

I have a DFI UT P45-T2RS. In the options, I can change the "SB/CPU PLL voltage."

Will a SB/CPU PLL increase from 1.55v (default) to 1.75v kill my CPU? I ask because I was aiming for 550FSB (or at least above 500FSB if not doable) and the jump in PLL has made it possible to do at low voltages for everything else. Can I keep it at 1.75v or am I risking major damage? I went from "not bootable" to "loads Windows and can run programs" with this minor increase. Still more tweaking to go, but it was a large jump....couldn't even get 500FSB stable before upping the PLL.

Voltages(Bios Values):
NB Core: 1.178v
VTT: 1.287v
SB/CPU PLL: 1.75v
Dram: 1.87v (They're made to run this low)
CPU: 1.2v

Edit: Ah crap, did I post in the right section?

[Kai Robinson]

Well, what CPU are you using? What memory?

[K404]

Yea. CPU matters for PLL. If its 45nm, id take it back to 1.65v and whack up the VTT. On P45, higher VTT is fine

[zsamz_]

i degraded 1 cpu with pl1,75;)
my e6850 was stable @ 4.2ghz n can only do 3.98ghz now
as soon as i hit 4ghz its not stable no more

[GrandMasterJ]

Sorry, I made a signature, don't know where it went. Specs are:

DFI UT P45-T2RS
Intel E8500 E0
Auzentech X-FI Prelude
4GB(2*2GB) G.Skill Pi series 1100Mhz
EVGA GTX 285 SSC
2x WD Raptor 74GB
2x Plextor 716AL DVD-Rom
Windows 7 RC
Enermax Modu 82+ 625watts

1.75v is the next available option. I can't set 1.65v. VTT may have to go up to 1.4 to bring stability in at 550FSB. Would that be a better option? I tested up to 1.32v and it seems stability is a far way out. Read about and tried upping the PLL to the next available option and everything worked...thing is...some people say it kills their CPU, some say theirs is fine after X months.

[GrandMasterJ]

i degraded 1 cpu with pl1,75;)
my e6850 was stable @ 4.2ghz n can only do 3.98ghz now
as soon as i hit 4ghz its not stable no more

How long did that take to happen?

[zsamz_]

How long did that take to happen?

a few months:mad:
woke up one day n it crashed @ 4.1 n then all i can get was 3.98
its still a great cpu as not many e6850 do 3.9
btw 45nm even more fragile with pll volts

[OC Nub]

I believe 1.57v is the max safe PLL for 45nm CPU's.

[GrandMasterJ]

Understood. Thanks for the replies. I brought it back down and I'll push VTT harder. I can pass two consecutive Intelburntests on "maximum" at 1.356v VTT and 1.198v NB... Going to run Prime blend from here to tweak GTL references and try to get ultimate stability. I can't seem to touch Performance Level at all...even pulling in one phase results in a C1...so, I'm pushing the FSB as much as I can to make up the difference.

I did not run the PC more than 20-30 minutes idle at 1.75v PLL while I waited for people to reply...no stressing. There's a possibility that, if that dangerous, I've degraded my CPU, but are chances unlikely?

Zsamz, once you pulled your PLL back, your chip has remained stable at what it has degraded to? No further degradation happens once the voltage is lowered?

[radaja]

maybe the vNB is a little on the low side for 500+fsb(1.17v-1.19v)might be the cause for the instabillity with 2x2GB kit?

[GrandMasterJ]

maybe the vNB is a little on the low side for 500+fsb(1.17v-1.19v)might be the cause for the instabillity with 2x2GB kit?

Hey radaja. Actually, it works with the above voltages (1.356v VTT and 1.198v NB)...Passes intelburntest consecutive runs. I need to run Prime Blend for awhile to see if there's any problems further down the road and find anything more to tweak.

Nice overclock, by the way. Now how the HELL did you get performance level 8 at 500FSB, please? :clap: What are your VTT and NB volts?

[zsamz_]

Understood. Thanks for the replies. I brought it back down and I'll push VTT harder. I can pass two consecutive Intelburntests on "maximum" at 1.356v VTT and 1.198v NB... Going to run Prime blend from here to tweak GTL references and try to get ultimate stability. I can't seem to touch Performance Level at all...even pulling in one phase results in a C1...so, I'm pushing the FSB as much as I can to make up the difference.

I did not run the PC more than 20-30 minutes idle at 1.75v PLL while I waited for people to reply...no stressing. There's a possibility that, if that dangerous, I've degraded my CPU, but are chances unlikely?

Zsamz, once you pulled your PLL back, your chip has remained stable at what it has degraded to? No further degradation happens once the voltage is lowered?


yes its stable now
i havent really played with it lately lol
i suggest you bump nb to 1.3
and no lol you didnt damage your cpu yet

[scharnhorst]

Hello,

Do you think that VPLL at 1.52v (on P5Q Deluxe) is safe to run a Q9650 H24?
Thank you in advance:yepp:

[WhiteFireDragon]

Hello,

Do you think that VPLL at 1.52v (on P5Q Deluxe) is safe to run a Q9650 H24?
Thank you in advance:yepp:

1.52v is fine, that's almost the lowest it can go

[scharnhorst]

All right thank you very much

[GrossmeisterB]

I'm still unsure, how long a high VTT-voltage will take to kill a 45nm-chip...I never will go above 1,30V on my P5Q-E, I think that's a safe value, but what do you guys think? Is it safe to give it some more? Or will it damage the chip soon?

[Leeghoofd]

Discussion is about PLL voltage sir G, but FYI there are loads of users that run over 1.35 FSB term daily (VTT voltage) without degredation or even killing a cpu. Some CPU's will just degrade ( very small minority ) even if you run them within Intel specs. If you don't feel comfortable stick within ya own safe zone, but you might miss out on a higher FSB.

With most of my 775 45nm Cpu's I never even went over 1.56 PLL voltage for 500-600FSB action. Think VTT is far more important as it also is related to the NB. Remember more isn't always better !!

[WhiteFireDragon]

Think VTT is far more important as it also is related to the NB. Remember more isn't always better !!

very true. sometimes when my vtt, pll, or vnb is too high, it actually causes a lot of instability. there's a sweet spot zone for all these voltages at certain clocks

[GrossmeisterB]

Discussion is about PLL voltage sir G, but FYI there are loads of users that run over 1.35 FSB term daily (VTT voltage) without degredation or even killing a cpu. Some CPU's will just degrade ( very small minority ) even if you run them within Intel specs. If you don't feel comfortable stick within ya own safe zone, but you might miss out on a higher FSB.

With most of my 775 45nm Cpu's I never even went over 1.56 PLL voltage for 500-600FSB action. Think VTT is far more important as it also is related to the NB. Remember more isn't always better !!

I know it was about PLL here, but I think there are still a lot of users having problems with PLL/VTT voltages etc.

I will try with a higher VTT and see if I can get the FSB any higher. But when I change the VTT-voltage, I will have to play with the GTLs again, right?

I also found out, that the PLL voltage doesn't matter that much, I tried 1.52, 1.54 and 1.56 and didn't see any differences...

Is it right (I read it somewhere), that VTT should be BELOW VNB to keep a good stability? And what's the max. VNB for the P45-chip? I had mine on 1.42 allready, I changed the stock heatpipe to an "Alpenföhn Ötzi" with a 80mm fan blowing through it, so there's enough cooling I think

[GrandMasterJ]

I know it was about PLL here, but I think there are still a lot of users having problems with PLL/VTT voltages etc.

I will try with a higher VTT and see if I can get the FSB any higher. But when I change the VTT-voltage, I will have to play with the GTLs again, right?

I also found out, that the PLL voltage doesn't matter that much, I tried 1.52, 1.54 and 1.56 and didn't see any differences...

Is it right (I read it somewhere), that VTT should be BELOW VNB to keep a good stability? And what's the max. VNB for the P45-chip? I had mine on 1.42 allready, I changed the stock heatpipe to an "Alpenföhn Ötzi" with a 80mm fan blowing through it, so there's enough cooling I think

The GTLs are set via a percentage. I do not think they have to be changed once the VTT voltage has been changed if set correctly, but somebody might say different. I've always adjusted my GTL refs after getting the VTT and NB to the correct level...the GTLs help bring that last bit of stability (for example: passing 24 hours Prime Blend instead of failing after 11 hours).

PLL did help my 550FSB, but after what I learned here and from kitfit at the dfi forum, I don't think I'll touch it again.

And yes, there is a "sweet spot"...I've had higher voltages actually fail where lower ones with correct GTLs are "100%" stable.

To Zsamz: I actually don't need that high of NB volts. It's up and runnin' at 1.198v Bios (smart guardian says 1.18v). I'm unsure if I want to push it higher to try to bring down performance level. I'm still below Intel's max value(1.2v) at this point. I may have gotten one of the higher clocking P45s.

But this has gone a bit off topic for the thread. End conclusion for future readers is: don't raise PLL voltage, it will most likely degrade your CPU

Can it cause degradation of the chipset as well? Or it only affects the CPU?

[K404]

vTT can be higher than vNB, I dont think thats a problem

vTT should be kept low on all chipsets other than P45. On P45 higher values are fine and quite often necessary for higher FSB

the NV volts are impressively low but slightly higher wont do any harm. Sometimes less isnt always cooler either! Depends on the PWM behaviour..though for an NB thats not so much of an issue :)

[knopflerbruce]

Sorry for OT, but what is considered max VTT during benchmarking on these Wolfdales?

Just had to ask, since VTT talk came up;)

[GrandMasterJ]

Sorry for OT, but what is considered max VTT during benchmarking on these Wolfdales?

Just had to ask, since VTT talk came up;)

I thought the max VTT was related to the NB chipset? Intel's quoted spec is 1.32v VTT for P45 (just checked the PDF. 1.21v on the NB). I've read some overclockers stating that 1.4v VTT is the "absolute max."

[knopflerbruce]

I thought the max VTT was related to the NB chipset? Intel's quoted spec is 1.32v VTT for P45 (just checked the PDF. 1.21v on the NB). I've read some overclockers stating that 1.4v VTT is the "absolute max."

Yeah, it's P45 (otherwise it would be completely OT:p:). Thanks.

[kitfit1]

Whatever Intel says the max VTT is, it really is a moot point anyway. If you check my OCDB Entry (http://csd.dficlub.org/forum/showpost.php?p=65013&postcount=15) you will see that i needed all of 1.2v for the VTT, that was at 563 fsb with 4x1gb sticks of ram. As has been said, sometimes less is more. Get the GTL's right, with the right amount of VTT and Vcore, and she's good to go.

[Ket]

I run 1.6v on my quad, I really don't recommend higher, even though arguably theres no "direct" evidence of PLL voltage degrading a CPU - thats down to user experience IMO.

[GrandMasterJ]

Whatever Intel says the max VTT is, it really is a moot point anyway. If you check my OCDB Entry (http://csd.dficlub.org/forum/showpost.php?p=65013&postcount=15) you will see that i needed all of 1.2v for the VTT, that was at 563 fsb with 4x1gb sticks of ram. As has been said, sometimes less is more. Get the GTL's right, with the right amount of VTT and Vcore, and she's good to go.

I took a different route. My VTT is higher (1.356v) and my NB is lower (1.198v). Is one option better/more recommended than the other? Still working on the GTLs.

Also, what do you mean by
Whatever Intel says the max VTT is, it really is a moot point anyway?