"ASUS P5Q Deluxe discussion".

[kiikkuja]

dang this thread moves so ing fast!...anyways

i have my 8400 stable at 4.5 with every thing on auto except my memory and cpu voltage why does everyone say that you have to cahgne the gtl and nb and pll i left everything on auto and i left occt running for 6 hrs on a cpu test and 3 hrs on blend so im guessing im 100% stable

Because the AUTO values overvolt like mad!! Seriously put your voltage settings couple notches above auto and you have nothing to worry about. The GTL values as everything else have a default setting which the BIOS should provide to you. Max values for PLL and fsb termination voltage are 1.60 and 1.38. You should never go above them atleast not 24/7 anyway.

And if you haven't run the most recent Prime95(V25.6) atleast 8 hours per large and small FFT:s your not stable imo. I just ran all the 3 tests available alteast 9 hours a piece. I haven't checked the IntelBurntest myself but I hear it's very good at finding problems with stability very fast.

PS. If you decide to use Prime95 you should have the round off checking setting enabled otherwise it's useless to run the program.

[Tulû]

Tried to raise dram from 2.14v to 2.16v ; got error after 15 mins again on core0 with blend in 2nd 8k test.
I tried to run a custom blend test after it and then i set 8k test manually and ran , there was no error in 2nd 8k test this time
I can't understand

Because the AUTO values overvolt like mad!! Seriously put your voltage settings couple notches above auto and you have nothing to worry about. The GTL values as everything else have a default setting which the BIOS should provide to you. Max values for PLL and fsb termination voltage are 1.60 and 1.38. You should never go above them atleast not 24/7 anyway.

And if you haven't run the most recent Prime95(V25.6) atleast 8 hours per large and small FFT:s your not stable imo. I just ran all the 3 tests available alteast 9 hours a piece. I haven't checked the IntelBurntest myself but I hear it's very good at finding problems with stability very fast.

PS. If you decide to use Prime95 you should have the round off checking setting enabled otherwise it's useless to run the program.

I ran prime small FFTs and intel burn test and it was stable ; but yesterday while i was palaying a game i got application error ( some mem error )
Then ran a blend test and got error on core0 .
I still get error only on core0 .
Btw , you've said that the max for FSB Term. is 1.38 .
I have it 1.48V
Pffffff I don't like oc anymore.
I just want a stable system

[CryptiK]

Tulu - run your ram slower, to take it out of the equation when overclocking your cpu. Doing both at once is bound to cause issues and confuse you, as you can't pin down the offending piece of hardware when there is instability.

Find max cpu clock, then adjust the ram to suit, as you will gain more from running cpu at max and having ram slower than having flying ram and a slower cpu.

[kiikkuja]

I ran prime small FFTs and intel burn test and it was stable ; but yesterday while i was palaying a game i got application error ( some mem error )
Then ran a blend test and got error on core0 .
I still get error only on core0 .
Btw , you've said that the max for FSB Term. is 1.38 .
I have it 1.48V
Pffffff I don't like oc anymore.
I just want a stable system

My advice was adressed to xTKxhom3r but should apply to you regarding the
stability testing. I may have been wrong on the voltage subject as it depends
on the processor used. My 45nm Q9450 is very sensitive but I can't say anything about your proc.

Leeghoofd please help us on the voltage subject regarding the q6600 and e8400 as I answered before I thought about it.

Edit.
PS. Reading through this thread would give surely the answer to most problems.
1+ on the Cryptik's suggestion BTW. I had my ram on AUTO or ~800-900mhz when trying to find max CPU freq.

[Mumid]

Well worth getting a DMM for testing voltages on these boards. I think all boards and even bios' are slightly different. A DMM can be had for peanuts these days.

[Tulû]

Tulu - run your ram slower, to take it out of the equation when overclocking your cpu. Doing both at once is bound to cause issues and confuse you, as you can't pin down the offending piece of hardware when there is instability.

Find max cpu clock, then adjust the ram to suit, as you will gain more from running cpu at max and having ram slower than having flying ram and a slower cpu.

My advice was adressed to xTKxhom3r but should apply to you regarding the
stability testing. I may have been wrong on the voltage subject as it depends
on the processor used. My 45nm Q9450 is very sensitive but I can't say anything about your proc.

Leeghoofd please help us on the voltage subject regarding the q6600 and e8400 as I answered before I thought about it.

Edit.
PS. Reading through this thread would give surely the answer to most problems.
1+ on the Cryptik's suggestion BTW. I had my ram on AUTO or ~800-900mhz when trying to find max CPU freq.

Okey,
i lowered my rams to 851MHz FSB : DRAM 1:1 now.
I ran prime , i got error again on core0 just 5 mins. after this time.
I just want to know that , i 've ran small FFT's 3 hours and 5 rounds of Intel burn test it was stable ; i wonder that if passes smallFFTs and IBT ,couldn't i say cpu speed stable ?

I've ran as 851MHz @ 5-5-5-18 trfc 55 @ 2.1 V

Edit: Now lowered FSB to 413MHz ( 413 x 8 ) and set rams to 826MHz @ 5-5-5-18 trfc 55 @2.1 V

Btw kiikkuja, you've said "Max values for PLL and fsb termination voltage are 1.60 and 1.38." but my FSB Term is 1.48 now and PLL is 1.58 .
This FSB Term. can cause a problem ?

[Polizei]

Max values for PLL and fsb termination voltage are 1.60 and 1.38. You should never go above them atleast not 24/7 anyway.

Is that for 45nm or 65nm or both?

[truehighroller]

Is that for 45nm or 65nm or both?

There you are did you ever test your stuff with those settings changed?

[kiikkuja]

Okey,
i lowered my rams to 851MHz FSB : DRAM 1:1 now.
I ran prime , i got error again on core0 just 5 mins. after this time.
I just want to know that , i 've ran small FFT's 3 hours and 5 rounds of Intel burn test it was stable ; i wonder that if passes smallFFTs and IBT ,couldn't i say cpu speed stable ?

I've ran as 851MHz @ 5-5-5-18 trfc 55 @ 2.1 V

Edit: Now lowered FSB to 413MHz ( 413 x 8 ) and set rams to 826MHz @ 5-5-5-18 trfc 55 @2.1 V

Btw kiikkuja, you've said "Max values for PLL and fsb termination voltage are 1.60 and 1.38." but my FSB Term is 1.48 now and PLL is 1.58 .
This FSB Term. can cause a problem ?

As I stated I'm not at all sure about the voltages anymore, because you have a different processor than mine and this is my first Intel platform I've ever overclocked. With Q9450 the values I stated are those but they may not be the ones that apply to your case.

I AM NOT AT ALL SURE ABOUT THE VOLTAGE ISSUE SO DON'T TAKE MY WORD ON IT.
Just to be sure you could lower your fsb termination voltage to under 1.40, ~1.38 maybe and if everybody agrees you can raise it when your sure it's safe. That way you don't break anything and enjoy the comp little bit longer.

On the core 0 issue you could try raising your vcore or gtl reference voltage for that core a notch or two.

The stability in my opinion is a matter of own preference. In my case I ran Prime 95 Large FFT for 21 hours, small FFT for ~10 hours and blend for ~9 hours just to be sure that when I compile programs in Gentoo there's no error
resulting from my overclock. If you consider you have not some critical application that needs absolute staibility then why worry so much? If you do
you should use all 3 programs (IntelBurntest, Prime95, OCCT) IMHO to be sure
the rig is completely stable at all times. And not some 1 hour this and 2 hours of that but rather 8 hours or overnight etc...

Although I ran Prime95 for a very long time it seems that my comp isn't absolutely 3d stable after all. When gaming under wine in Gentoo cod2 stops at some point with sound on but nothing happens on my screen.
I had lower my clocks and vcore to be able to play properly the game.
I'm quite sure that my radiator is too small and cannot cool this quadcore monster and a overclocked 8800GT at the same time.

[kiikkuja]

Is that for 45nm or 65nm or both?

To my knowledge they apply to 45nm. I've been told that the older 65nm aren't
so fragile.

[Tulû]

As I stated I'm not at all sure about the voltages anymore, because you have a different processor than mine and this is my first Intel platform I've ever overclocked. With Q9450 the values I stated are those but they may not be the ones that apply to your case.

I AM NOT AT ALL SURE ABOUT THE VOLTAGE ISSUE SO DON'T TAKE MY WORD ON IT.
Just to be sure you could lower your fsb termination voltage to under 1.40, ~1.38 maybe and if everybody agrees you can raise it when your sure it's safe. That way you don't break anything and enjoy the comp little bit longer.

On the core 0 issue you could try raising your vcore or gtl reference voltage for that core a notch or two.

The stability in my opinion is a matter of own preference. In my case I ran Prime 95 Large FFT for 21 hours, small FFT for ~10 hours and blend for ~9 hours just to be sure that when I compile programs in Gentoo there's no error
resulting from my overclock. If you consider you have not some critical application that needs absolute staibility then why worry so much? If you do
you should use all 3 programs (IntelBurntest, Prime95, OCCT) IMHO to be sure
the rig is completely stable at all times. And not some 1 hour this and 2 hours of that but rather 8 hours or overnight etc...

Although I ran Prime95 for a very long time it seems that my comp isn't absolutely 3d stable after all. When gaming under wine in Gentoo cod2 stops at some point with sound on but nothing happens on my screen.
I had lower my clocks and vcore to be able to play properly the game.
I'm quite sure that my radiator is too small and cannot cool this quadcore monster and a overclocked 8800GT at the same time.

I see
Here is the good news ; i 've lowered the FSB to 413MHz and now it's stable for about 70 mins.
It seems that 1.37500 Vcore was low for 425MHZ FSB.
Now , should i go on to run blend or stop it and run a small FFTs ; so what is my next step ?
Thanks to all of you.

[CryptiK]

Tulu, try this:

FSB : 425 x 8
PCIE: 100
DRAM : 851 // 5-5-5-15 (trfc:40) @2.1v
DRAM STATiC READ.. : DISABLED
VCORE: 1.38v
CPU PLL : 1.58
GTL REF (0/2, 1/3) = 0.67x on both
NB GTL REF = AUTO
FSB TERM. : 1.46
DRAM: 2.1
NB: 1.34
SB: AUTO
PCIE SATA: AUTO
LLC: ENABLED

If it fails, set cpu GTL Ref (0/2, 1/3) = 0.69x and try again

[RAdoxx]

Does every voltage setting overvolt? Or only vDIM, vFSB, vPLL and vNB like in the picture posted a few pages back?

[CryptiK]

No definitive answer on that yet, we know the measure points for the SB but I haven't pulled my board out and measured it.

[jas420221]

Here are my 24/7 stable settings for Q9450 @ 3.8Ghz **updated from prior posts**

Vcore= 1.34
Vcoregtl=.68/.68
FSBt= 1.32
PLL= 1.54
NB= 1.32
NBgtl= .68

[Tulû]

Tulu, try this:

FSB : 425 x 8
PCIE: 100
DRAM : 851 // 5-5-5-15 (trfc:40) @2.1v
DRAM STATiC READ.. : DISABLED
VCORE: 1.38v
CPU PLL : 1.58
GTL REF (0/2, 1/3) = 0.67x on both
NB GTL REF = AUTO
FSB TERM. : 1.46
DRAM: 2.1
NB: 1.34
SB: AUTO
PCIE SATA: AUTO
LLC: ENABLED

If it fails, set cpu GTL Ref (0/2, 1/3) = 0.69x and try again

Tried and failed on core0 , will try GTL Ref ( 0/2 , 1/3 ) = 0.690 now.

[CryptiK]

Also try increasing vcore to 1.40 - 1.42v (bios) it may just be your cpu. You should be safe up to at least 1.45v vcore with the 65nm if you can keep the heat under control.

[Tulû]

Yes, i know these limits for 65nms .
But the weather is too hot here , my room is appr. 30C now.
My cores are max: 67 67 61 62 with realtemp with 1.38125V for now . ( at that moment it makes 8k test with blend )
I know that these temps won't be so high anytime except torturing.
There is another question btw , what are the safe temps for torturing,bench in realtemp ?

Btw , stable for about 25 mins. ( with GTL REFs 0/2 , 1/3 = 0.690 )

[RejZoR]

Old 65nm chips are safe up to 1,55V with good air cooling or 1.5V with usual decent aftermarket cooling.

[Tulû]

Also try increasing vcore to 1.40 - 1.42v (bios) it may just be your cpu. You should be safe up to at least 1.45v vcore with the 65nm if you can keep the heat under control.

Failed on core0 after one hour.
Will increase vcore a notch.

[CryptiK]

Yes, i know these limits for 65nms .
But the weather is too hot here , my room is appr. 30C now.
My cores are max: 67 67 61 62 with realtemp with 1.38125V for now . ( at that moment it makes 8k test with blend )
I know that these temps won't be so high anytime except torturing.
There is another question btw , what are the safe temps for torturing,bench in realtemp ?

Btw , stable for about 25 mins. ( with GTL REFs 0/2 , 1/3 = 0.690 )

Fair enough about the temps, that's why I mentioned it, not knowing where you live or your climate.

Hope the GTL REF gets you that FSB stable. Many people even run 0.70x (0/2, 1/3) on the 6600's. Also try reducing the FSB term voltage to around 1.42v an using 0.70x - 0.72x GTL REF. Lowering the vFSB will reduce your temps slightly.

[Polizei]

There you are did you ever test your stuff with those settings changed?

Yeah, it helped. I got 378-380FSB stable at 1.5125v BIOS (2 notches above 1.5) for 1.504v load according to CPU-Z. It's got me 3.4GHz, but I can't seem to go any higher. If I set a higher FSB or change my RAM speed in BIOS, when I save and exit, it won't POST. I flip the switch on the power supply, reboot and then it tells me that my OC failed. Once I'm in Windows, if I do a shutdown or restart through Windows, it reboots fine. Warm reboot from BIOS though, won't work.

I was hoping my 975x chipset was holding me back, but it seems my B2 E6600 is not a good clocker...

I want to OC my RAM now and tighten the timings, but I think it's on it's way out and dieing, so I can't get very far. I increase the voltage, tighten the timings slightly (4-4-4-12 @ 800mhz @ 2.16v instead of 5-5-5-15 @ 757mhz @ 1.98) and it won't boot. I KNOW this RAM is capable of doing that easily because I did it on my old board. It would do 950mhz 4-4-4-10 @ 2.2v all day stable on my old 975x DFI.

Don't know what to do... 333 and 400 strap change nothing.

[Tulû]

Fair enough about the temps, that's why I mentioned it, not knowing where you live or your climate.

Hope the GTL REF gets you that FSB stable. Many people even run 0.70x (0/2, 1/3) on the 6600's. Also try reducing the FSB term voltage to around 1.42v an using 0.70x - 0.72x GTL REF. Lowering the vFSB will reduce your temps slightly.

I follow you .
Did not increase vcore again , hold it at 1.38125.
Lowered FSB term. to 1.42
GTL REF 0/2 = 0.710
GTL REF 1/3 = 0.720


Here are my all settings :

 AI Overclock tuner: manual
CPU Ratio Setting: 8
FSB Strap to North Bridge: AUTO
FSB Frequency: 425
PCI-E Frequency: 100
DRAM Frequency: 851
DRAM CLK Skew on Channel A1: AUTO
DRAM CLK Skew on Channel A2: AUTO
DRAM CLK Skew on Channel B1: AUTO
DRAM CLK Skew on Channel B2: AUTO
DRAM Timing Control: manual

1st Information : 

CAS# Latency: 5
DRAM RAS# to CAS# Delay: 5
DRAM RAS# Precharge: 5
DRAM RAS# Activate to Precharge: 15
RAS# to RAS# Delay : AUTO
Row Refresh Cycle Time: 40
Write Recovery Time: AUTO
Read to Precharge Time: AUTO

2nd Information :

READ to WRITE Delay (S/D): AUTO
Write to Read Delay (S): AUTO
WRITE to READ Delay (D): AUTO
READ to READ Delay (S): AUTO
READ to READ Delay (D): AUTO
WRITE to WRITE Delay (S): AUTO
WRITE to WRITE Delay (D): AUTO

3rd Information :

WRITE to PRE Delay: AUTO
READ to PRE Delay: AUTO
PRE to PRE Delay: AUTO
ALL PRE to ACT Delay: AUTO
ALL PRE to REF Delay: AUTO
DRAM Static Read Control: Disabled
DRAM Read Training: enabled
MEM. OC Charger: enabled
AI Clock Twister: lighter
AI Transaction Booster: AUTO

CPU Voltage: 1,38125
CPU GTL Voltage Reference (0/2): 0.710
CPU GTL Voltage Reference (1/3): 0.720
CPU PLL Voltage: 1.52
FSB Termination Voltage: 1.42
DRAM Voltage: 2.10
NB Voltage: 1.34
NB GTL Reference: AUTO
SBridge Voltage: AUTO
PCIE SATA Voltage: AUTO

Load Line Calibration: enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew : auto
NB Clock Skew : auto

Advance CPU Settings
CPU Ratio Setting: Manual
CPU VID: Default
C1E Suppport: Disabled
Max CPUID Value Limit: Disabled
Intel® Virtualization Tech: Disabled
Vanderpool Technology: Disabled
CPU TM Function: enabled
Execute Disable Bit: Disabled

Btw , i live in Turkey .
Harry Kewell plays for us .

Edit :Ooopss , system restarted.
It seems Fsb term 1.42 was low , what do you think ?
I increased to 1.46 again.

Edit2: Restarted again with 1.46 FSB TERm.
Seems GTL REFs high ?
Lowered to 0.700 both.

[CryptiK]

Yeah perhaps it was too low, try 1.46 again with 0.70x GTL refs & see how that goes. How did it go with 1.46 vFSB and 0.69x GTL Refs?

Damn you for recruiting Kewell he was one of our best players!

EDIT - just saw your edit - if it's always core 0 that fails, try using 0.70x for 0/2 and 0.69x for 1/3.

[Tulû]

It failed on core0 after one hour with 0.690 GTL REFs and 1.42 FSB TERM.
Yes, he is a great player sure.

I like his personality too much. He is a gentleman first.
He scored on this sunday in super cup and this was his first official goal.
He was sub. and he joined the game at 65. min. and he scored at 66. min

Edit: I just saw your edit ,
now , they both 0.70 and FSB TERM 1.42 ,passed 20 mins and no restart.