ASUS Maximus II Formula - new P45 king?

[64dragon]

everything i tried with tuning the GTL's made me less stable,

tried:
-5,-45,-5,-45 and that lasted 19minutes in Orthos (core 4 failed)
10,-35,10,-35 lasted 9 minutes (core 4 failed)
i looked at the calculation again and saw that 10,-30,10,-30 was a closer match in voltage but that lasted 6 minutes (core 4 failed)
it seemed going this way was worse so i tried going the other way
-10,-55,-10,-55 lasted 5 minutes (core 1 failed), that threw a wrench in my theory so now i have no clue what to try
then i put the GTLs back on Auto and raised VTT and vNb 2 notches each and that lasted 1hr 30min (core 4 failed)

i really don't know what to do, this chip is pissing me off. this one might be going to my dad unless it clocks 10x better on a UD3 that i plan on getting for his pc (since the board is cheap and blue would clash with the red theme i have going)

edit: is there a software program that will display what the GTL values are in OS? if i were to set VTT and the GTLs back to auto and become stable, how could i find what BIOS was setting them to?

[humeyboy]

Have you tried Positive Values ?

[64dragon]

Have you tried Positive Values ?

not yet. using the equation, i got negative values so i stayed with that

[CryptiK]

AFAIK there's no app that can read the GTL's in windows. The manual doesn't exactly say what auto sets the Gtl's to and if this changes with an OC, so its anyones guess.

I would recommend you try positive values, the equation is to give people a starting point for further tweaking. Its also good for those that dont know how/dont have the inclination to tweak, but want to set the GTL's to 'something other than auto'.

You should just see what works best for your cpu and ignore what the equation says, as its not giving you the results you're after.

Try something like +50/+10/+50/+10 or higher even with Vtt set to ~1.20v for starters, perhaps increase Vtt to ~1.25v also and see what affect that has.

[humeyboy]

not yet. using the equation, i got negative values so i stayed with that

The Calcs are there as a starting point, you just 9 times out of ten make it Neg, I had to use Pos to get 475FSB.

My NB will not boot at Neg Values (well at overclocked speed) so its also Pos.

[In-Fluence]

Although the board could probably handle PL6 at 450 FSB, it depends on the ram divider. The 266 strap will also require more voltage than the 333 so it might require a bump to post and a further increase to be stable.

I am running a Rampage Extreme currently, and I wouldn't advise you to pair it with a Q9650. From other guys results, you will probably struggle to get anything over ~475 FSB stable with the quad without using a lot of vNB. However if that's all you want out of it it will be fine. The REX is much better paired with a wolfdale if you want a flexible range of usable FSB. If you want the most out of a s775 quad, the UD3P/R really seem to be where its at.

I run Corsair 1866 7-8-7-20 (Two 2GB sticks of the 6GB triple channel kit) Elpida Hyper based ram on the REX. They are good, but if you want to push high MHz with tight timings and plenty of vdimm, a 2 x 1GB Micron D9 GTR/GTS kit is what you want.

Yep, PL6 was a no-go, even going from PL7, stronger which was 10hours blend stable, to PL6, lighter wouldn't post. I'm happy to settle for that speed for now and can probably tweak a few more settings before calling it a day on that.
As for the REX, thanks for the advice. I've noticed the Gigabytes have been doing well in this area too so I will keep an eye on those. I quite liked the DFI LP UT NF4 SLI DR expert (or whatever it was) in the past, but it was very tricky to tweak, and so I am slightly dubious. DFI's warranty policy, sounds pretty strict from what I've heard too, but they've got the appealing option of being extremely tweakale.
Anyway, I may well still order a Gigabyte & 9650 and do a cheeky swap for my Q6600 to see how the 45nm quad performs on here, and maybe the mobo if it seems to do any better, but I do want to keep temps as low as possible for a near to silent system which makes me think the MIIF might be better. The whole DDR3 thing isn't quite so important for me right now and will probably try and stick with some combination of the aforementioned before I consider a more major upgrade in a year or two, but it would be interesting to get a feel for it.

Now payday's here, I've ordered a Corsair HX850, a Swiftech NB-Max block and 2 more feet of tygon tubing so I can sell off the MCW30 before it depreciates in value and hopefully improve the chipset/PWM cooling at little cost (and finally get rid of that awful bit of blue XSPC tubing and inefficient PSU). Good old tekheads have some great stock

One final thing (sorry to rant), now that I found a nice stable point for my system I tried lowering the vcore and re-tuning the GTLs, which have now made me think that they were'nt done properly (again)! I lowered the vcore to 1.425v (-.025v) whilst staying at 450x8 to make it unstable.

I tested with current GTLs (that I thought were the most stable) 50,10,50,10 and it reset in 2-3 minutes.
Then went 10,10,10,10 and 1024k paseed, and core 4 failed after 12 minutes of 8K (and gave a rounding error)
Going to 20,20,20,20 made 8K fail in about 2-3 minutes (followed by a reset)
So I tried -30,10,-30,10 as I had a hunch that perhaps GTLs 1 and 3 might be correctly tuned, and 0 and 2 were off and it passed 8K but almost immediately failed 10k and restarted
At -35,10,-35,10 10K ran for about 3 minutes longer before it restarted (but is the best so far)
Tested -25,20,-25,20 and it hung within 1 minute of 8K
Now testing -45,0,-45,0 (assuming auto is 0).
This time around, the results I'm getting from tuning the CPU GTLs feels 'right' whereas before results seemed a lot more random and no patterns were emerging (although I'm still rarely getting rounding errors, just reboots which still seems a bit off to me). Hopefully if I find one that stands out a lot more, I can start to push the CPU again, but I'm not budging until I'm certain of the best GTLs for me this time!

[humeyboy]

I don't get why you would need waste cash on WC for the Mobo.

They run far cooler (if not launch models, as Asus sorted heat issue with badyl mounted Heatpipes or you refit Heatpipes/TIM yourself) than say a Nvidia Chipset Mobo and hardly need any voltage to get to about 475FSB (1.26-1.28v NB) on a Quad and less for a Dual.

My NB and SB are about 43C and never go over 52C even in the summer with all fans at 50% and Priming or Intel Test Burning for 24 hours.

[In-Fluence]

I had the MCW30 block in the loop left over from my P5B setup and it seemed to help out on temps a reasonable amount once I managed to get a portion of the block on the mounting area (certainly did no harm).
I'm still having issues with the mosfet temps which seem to get worse when going beyond 1.45vcore and I normally get a reboot when they get to around 70°C. Using enzotech mosfet sinks without the heatpipe assembly resulted in even higher temps (all overheat protection bios options are disabled too), but both the NB & SB were noticeably cooler. I now have the heatpipe assembly back on it and it runs stable with the case open & house fan on it, so I'm hoping the NB-Max block will help out with this problem a little more by inducting that little bit more heat from the mosfet area, maybe not a great deal but it's of little cost and gives me peace of mind (unless I get a mosfet block).

I'm on 1.41v NB at the moment so it does run a little warm, but will try lowering it once I've got the CPU GTLs tuned.
As for those little critters, I've just tried -40,+10,-40,+10 and I got 1024k, 8k and 11 mins of 10k before a reboot, the best so far by about 8 minutes, but a gap of 50mv?!?!
Now trying -45,+10,-45,+10 to see if furthering this gap to 55mv offers any improvements - perhaps unorthodox going by the equation, but who knows?!
Forgot to mention I'm only running 1.32 vtt throughout all these tests so it would seem that 40/45mv should be the best gap.

[Haes]

not yet. using the equation, i got negative values so i stayed with that

no it's not negative values, cause you've to think in mathematical logic:

- * + = -

- * - = +

+ * + = +

so vtt*GTLref1-value=vtt*GTLref2

for example:

VTT=1.20 V

ideal GTLvalue 0.667% (or 0.67x)...

* = Quadcore needs a bit more max 0.7%
^ =Dualcore needs a bit less max. 0.635%

looking for idealvalue...

GTLref1= * for Quad, ^ for Dual
GTLref2 = ideal GTLvalue

1.2V*0.7x - value = 1.2V*0.667x

value = 1.2*0.7-1.2*0.667 = 1.2*(0.7-0.667) = +33 mV Quadcore
.................................................. ........... = -32mV Dualcore

[64dragon]

no it's not negative values, cause you've to think in mathematical logic:

- * + = -

- * - = +

+ * + = +

so vtt*GTLref1-value=vtt*GTLref2

for example:

VTT=1.20 V

ideal GTLvalue 0.667% (or 0.67x)...

* = Quadcore needs a bit more max 0.7%
^ =Dualcore needs a bit less max. 0.635%

looking for idealvalue...

GTLref1= * for Quad, ^ for Dual
GTLref2 = ideal GTLvalue

1.2V*0.7x - value = 1.2V*0.667x

value = 1.2*0.7-1.2*0.667 = 1.2*(0.7-0.667) = +33 mV Quadcore
.................................................. ........... = -32mV Dualcore

your equation is wrong, in bios the equation and using your #'s would be 1.2V*0.7x + value = 1.2V*0.667x
value = 1.2V*0.667x - 1.2V*0.7x = 1.2V*(0.667-0.7) = -.0396v = -40mv
i've been pluging in the equation to a ti-89 calculator and having it solve for my values...until i made my own excel sheet

either way, i tried:

50,10,50,10 and failed in 10 minutes on core 4
70,30,70,30 failed in 21 minutes on core 4 (my longes time up to this point)
80,40,80,40 failed in 5 minutes on core 4 (a bit worse than above run)
90,50,90,50 failed after 1hr 13min on core 4 (new record for longest run)
and just failed at 100,60,100,60 at 41 min on core 4 but with a fatal error instead of a rounding error like the rest. not sure what that means in relation to GTL tuning

so, do i go higher to 110,70,110,70 to see what happens or go back to 90,50,90,50 and then up my VTT and vNB until i'm stable? How high of a GTL value is safe? since i'm at .94v
my Xeon was 10 times easier to OC than this...but may have been able to lower voltages had i tried tuning the GTLs then.

[humeyboy]

See all that BS info in the Calc !, it is not even needed as it only confuses most peeps.


Example Quoted from elsewhere but far simpler than this sites :

FSB Termination: 1.35v
Formula:
Vtt x 0.667 + Item = Vtt x 0.635
1.35v x 0.667 + Item = 1.35v x 0.635
0.90045v + Item = 0.85725v
Item = 0.85725 – 0.90045
Item = - 0.0432v
Item= -43mv


Now round up to the nearest selectable value in the bios is -40mv and set the 0.667x Item value
to -40mv (negative), usually the (1 &3) core.
You have now made the 0.667x GTL voltages as close as possible to the 0.635x GTL voltages.

Bios setting Example:
CPU GTL REFERENCE (0) Auto
CPU GTL REFERENCE (1) -40mv
CPU GTL REFERENCE (2) Auto
CPU GTL REFERENCE (3) -40mv


Notice:
If the negative value is unstable then you might have to try the positive value and you also have to
tweaking it by trial and error until you find the stable value for your processor.



Now my way without Algebra crap.


Example :

FSB Term Voltage (VTT) = 1.2v

1) 1.2v x 0.667 = 0.8004

2) 1.2v x 0.635 = 0.762

3) 0.8004 - 0.762 = 0.0384 or 0.762 - 0.8004 = -0.0384

4) GTL's are +/- 0.38v so closest in our Bios is +0.40 or -0.35 or -0.40v (Neg are in 0.50v Steps, Pos are 0.10v in steps).

[Haes]

strange, that no matter which quad i had , all works great with "+" prime stable...

my Q6600 G0 30 10 30 10 30 3,6Ghz 1,343VC
my Q9550 E0 50 10 50 10 30 ~4,0Ghz 1,343VC
my Q9650 E0 20 10 20 10 30 4,2Ghz 1,328VC

could decrease 2-3 steps in VC everytime

and +30mv nb gtl (like already mentioned above) best stable NB for my in every fsb frequency

FSB VTT Temp

450 1.15 49°C
471 1.22 51°C
490 1.33 54°C max rock stable FSB
500 1.36 54°C bootable 15min stable
515 1.38 55°C bootable only 2 Cores, 3min stable

all with 2*4 gb RAM

nothing washed, all stock 1.02G A2 already about one year old


seriously, the only thing that bothers me, is the great Vdrop 0.08-0.09V.

[CryptiK]

@ Influence - There is a couple of DFI P45 boards that look like they perform ok with quads, but the best results I have seen are consistently on the UD3P or UD3R. DFI's are tweakable, but this does not always equate to better FSB or better performance. My DK X48-T2RSB Plus is a great board, especially for ram clocking (which is what I bought it for), but there are other X48 DDR2 boards that can achieve higher FSB etc. Depends what you want out of the board. There's plenty of good results in the UD3P thread and lots of guys using it, so getting advice & comparing settings is much easier than with the less popular DFI's.

@ Humeyboy - that equation is the same as the one I made in my GTL thread, just instead of 'item' I used 'Y'.

Here's one of mine for comparison (using Vtt of 1.14v):

Vtt x 0.667 + Y = Vtt x 0.635

1.14v x 0.667 + Y = 1.14v x 0.635
0.760v + Y = 0.724v
Y = 0.724v - 0.760v
Y = -0.036v
Y = -36 mv

@ 64dragon - you can always set the GTL's to the same mv offset too, like +50/+50/+50/+50, or decrease the gap, like +50/+40/+50/+40.

[humeyboy]

Yes but the extra non needed Algebra is BS and confuses many peeps even while at school (20+yrs ago) I remember most could not grasp it.

No need to make anything harder than it needs to be.

Half can be totally skipped, sure it looks good on paper but its confusing to most.

[In-Fluence]

Well, I think for the price, Gigabyte UD3P range sounds appealing. I will research further.

The longest stable CPU GTLs I found were at -55,-5,-55,-5 which still doesn't seem right, but narrowing the gap gave me less stability (faill during 8K), while keping it at 50mv aways seemed to get into the 10k tests. Either way, it's still a big improvement over 50,10,50,10 which failed during 1024k, and I suspect I will be able to drop the vcore down by 2-3 notches.
One thing I did notice that seems to make me question if this is the right setting is that when I tested at +10,+10,+10,+10 and -20,-20,-20,-20, they gave me a rounding error, all others just rebooted (no BSOD either)

[CryptiK]

Yes but the extra non needed Algebra is BS and confuses many peeps even while at school (20+yrs ago) I remember most could not grasp it.

No need to make anything harder than it needs to be.

Half can be totally skipped, sure it looks good on paper but its confusing to most.

I'm not sure how using the letter 'Y' as opposed to 'item' makes anything any harder to grasp, nor so I think it looks any better. You supplied an identical equation (maths wise) and described it as being 'way simpler than the one on this site' when the only difference is that it uses 'item' instead of 'Y'.

The whole equation working is there to show the process, of course once you get it you can skip whatever you want.

Well, I think for the price, Gigabyte UD3P range sounds appealing. I will research further.

The longest stable CPU GTLs I found were at -55,-5,-55,-5 which still doesn't seem right, but narrowing the gap gave me less stability (faill during 8K), while keping it at 50mv aways seemed to get into the 10k tests. Either way, it's still a big improvement over 50,10,50,10 which failed during 1024k, and I suspect I will be able to drop the vcore down by 2-3 notches.
One thing I did notice that seems to make me question if this is the right setting is that when I tested at +10,+10,+10,+10 and -20,-20,-20,-20, they gave me a rounding error, all others just rebooted (no BSOD either)

In my experience/opinion a reboot is less stable than a rounding error.

I'm not sure if a fatal error means its less stable than a rounding error. It sounds worse but who knows for sure.

Sounds like you need more vcore maybe more Vtt and just get it stable, then tweak the GTL's and try and lower vcore/vtt from that point.

[64dragon]

Hey guys, I think i'm acutally stable i made the below settings last night and woke up this morning to a 14.5hr large FFT pass. clearly, ram was never my issue cause i'm finally running it OCed. so, my question now, what is strap and how does it affect things? I went from the 333 to 400 and became stable. Does strap affect "rated FSB"?

Processor: Q9505
BIOS revision: 1802
Ai Overclock Tuner: Manual
Ratio CMOS Setting: 8.5
FSB Frequency: 409
CPU Clock Skew: Normal
NB Clock Skew: Normal
FSB Strap to North Bridge: 400
DRAM Frequency: DDR2-1090MHz

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

CAS# Latency: 5 DRAM Clocks
DRAM RAS# to CAS# Delay: 5 DRAM Clocks
DRAM RAS# Precharge: 5 DRAM Clocks
DRAM RAS# Activate to Precharge: 18 DRAM Clocks
RAS# to RAS# Delay: 3
Row Refresh Recycle Time: 55 DRAM Clocks
Write Recovery Time: 6
Read to Precharge Time: 3

Read to Write Delay (S/D): 8
Write to Read Delay (S): 3
Write to Read Delay (D): 5
Read to Read Delay (S): 4
Read to Read Delay (D): 6
Write to Write Delay (S): 4
Write to Write Delay (D): 6

Write to PRE Delay: 14
Read to PRE Delay: 5
PRE to PRE Delay: 1
All PRE to ACT Delay: 6
All PRE to REF Delay: 6
DRAM Static Read Control: Disabled
DRAM Read Training: Disabled
MEM. OC Charger: enabled
Ai Clock Twister: Moderate
Ai Transaction Booster: Manual

Common Performance Level: 12
Pull-in of CHA PH1: Disabled
Pull-in of CHA PH2: Disabled
...
Pull-in of CHB PH1: Disabled
Pull-in of CHB PH2: Disabled
...

PCIE Frequency: 101

CPU Voltage: 1.31
CPU PLL Voltage: 1.54
FSB Termination Voltage: 1.3385
DRAM Voltage: 2.118
North Bridge Voltage: 1.378
South Bridge 1.5 Voltage: 1.51325
South Bridge 1.1 Voltage: 1.11325

CPU GTL Reference (0): 90
CPU GTL Reference (1): 50
CPU GTL Reference (2): 90
CPU GTL Reference (3): 50
NB GTL Reference: Auto
DDR2 ChA Reference Voltage: Auto
DDR2 ChB Reference Voltage: Auto
North Bridge DDR Reference: Auto

CPU Configuration:

Ratio CMOS Setting: 8.5
C1E Support: Disabled
Max CPUID Value Limit: Disabled
Intel Virtualization Tech: Disabled
CPU TM Function: Disabled
Execute Disable Bit: Disabled

Load-Line Calibration: Disabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled

[humeyboy]

The FSB Strap gives you Memory Ratios to whatever FSB you set but it also changes Timings.

So 400 is slacker than 333 so your Memory Bandwidth will be lower but its easier on the NB.

Install latest Bios FFS.

Then set to 400FSB, 333 Strap and Memory either Auto or 400 and Moderate + PL 10 for 1:1.

It should work with very little voltages.

[Mr.Pink]

Hey guys, I think i'm acutally stable i made the below settings last night and woke up this morning to a 14.5hr large FFT pass. clearly, ram was never my issue cause i'm finally running it OCed. so, my question now, what is strap and how does it affect things? I went from the 333 to 400 and became stable. Does strap affect "rated FSB"?

Congrats on the stable settings...
There are much better ram settings than what your running, PL12 is excessive for that clock/strap.

[Haes]

@mr pink:

can you give your bios settings for 4.2ghz 9650 pls. I need to compare something... thx

[64dragon]

my "stability" deceived me, woke up this morning to a BSOD and 2 work units in WCG had computation errors, could be the ram since it is now OCed

[Haes]

My PC is empty at the moment since I'm waiting for my Q9650 to arrive so I took the opportunity to remove the motherboard heatsink and replace the crap that ASUS put on with my Zalman thermal paste. I was getting NB/SB temps of 45-48C with everything at stock so hopefully that'll improve. What sort of chipset temps are you guys getting, particularly those of you who have replaced the TIM with something better?

Lenny, can you give your bios settings too pls.

[Mr.Pink]

@mr pink:

can you give your bios settings for 4.2ghz 9650 pls. I need to compare something... thx

Certainly

Extreme Tweaker
Ai Overclock Tuner : Manual
OC From CPU Level Up : AUTO
CPU Ratio Control : Manual
- Ratio CMOS Setting : 9
FSB Frequency : 467
CPU Clock Skew : Delay 100ps
North Bridge Clock Skew : Delay 100ps
FSB Strap to North Bridge : 333
DRAM Frequency: DDR2-1121
Dram Clock Skew CH1 A1 : Normal
DRAM Clock Skew CH1 A2 : Normal
Dram Clock Skew CH1 B1 : Normal
Dram Clock Skew CH1 B2 : Normal


DRAM Timing Control: Manual
CAS# Latency : 5
RAS# to CAS# Delay : 5
RAS# Precharge : 5
RAS# ActivateTime : 15
RAS# to RAS# Delay : 3
Row Refresh Cycle Time : 55
Write Recovery Time : 6
Read to Precharge Time : 3

Read to Write Delay (S/D) : 8
Write to Read Delay (S) : 3
Write to Read Delay (D) : 5
Read to Read Delay (S) : 4
Read to Read Delay (D) : 6
Write to Write Delay (S) : 4
Write to Write Delay (D) : 6

Write to PRE Delay : 14
Read to PRE Delay : 5
PRE to PRE Delay : 1
ALL PRE to ACT Delay : 5
ALL PRE to REF Delay : 5

DRAM Static Read Control: Enabled
Dram Read Training : Disabled
MEM OC Charger : Enabled
Ai Clock Twister : Stronger
Transaction Booster : Manual

Common Performance Level [8]

Pull-In of CHA PH1 Disabled
Pull-In of CHA PH2 Disabled
Pull-In of CHA PH3 Disabled
Pull-In of CHA PH4 Disabled
Pull-In of CHA PH5 Disabled
Pull-In of CHB PH1 Disabled
Pull-In of CHB PH2 Disabled
Pull-In of CHB PH3 Disabled
Pull-In of CHB PH4 Disabled
Pull-In of CHB PH5 Disabled

PCIE Frequency : 103

CPU Voltage : 1.325
CPU PLL Voltage : 1.59275
FSBT : 1.37825
DRAM Voltage : 1.9
North Bridge Voltage : 1.4108
South Bridge Voltage 1.5 : 1.55300
South Bridge Voltage 1.1 : 1.20600

CPU GTL Reference 0 : Auto
CPU GTL Reference 1 : -45mv
CPU GTL Reference 2 : Auto
CPU GTL Reference 3 : -45mv
North Bridge GTL Reference : AUTO
DDR2 Channel A REF Voltage : AUTO
DDR2 Channel B REF Voltage : AUTO
North Bridge DDR Reference : AUTO

Load Line Calabration : Enabled
CPU Sread Spectrum : Disabled
PCIE Spread Spectrum : Disabled

[Haes]

Certainly

Extreme Tweaker
Ai Overclock Tuner : Manual
OC From CPU Level Up : AUTO
CPU Ratio Control : Manual
- Ratio CMOS Setting : 9
FSB Frequency : 467
CPU Clock Skew : Delay 100ps
North Bridge Clock Skew : Delay 100ps
FSB Strap to North Bridge : 333
DRAM Frequency: DDR2-1121
Dram Clock Skew CH1 A1 : Normal
DRAM Clock Skew CH1 A2 : Normal
Dram Clock Skew CH1 B1 : Normal
Dram Clock Skew CH1 B2 : Normal


DRAM Timing Control: Manual
CAS# Latency : 5
RAS# to CAS# Delay : 5
RAS# Precharge : 5
RAS# ActivateTime : 15
RAS# to RAS# Delay : 3
Row Refresh Cycle Time : 55
Write Recovery Time : 6
Read to Precharge Time : 3

Read to Write Delay (S/D) : 8
Write to Read Delay (S) : 3
Write to Read Delay (D) : 5
Read to Read Delay (S) : 4
Read to Read Delay (D) : 6
Write to Write Delay (S) : 4
Write to Write Delay (D) : 6

Write to PRE Delay : 14
Read to PRE Delay : 5
PRE to PRE Delay : 1
ALL PRE to ACT Delay : 5
ALL PRE to REF Delay : 5

DRAM Static Read Control: Enabled
Dram Read Training : Disabled
MEM OC Charger : Enabled
Ai Clock Twister : Stronger
Transaction Booster : Manual

Common Performance Level [8]

Pull-In of CHA PH1 Disabled
Pull-In of CHA PH2 Disabled
Pull-In of CHA PH3 Disabled
Pull-In of CHA PH4 Disabled
Pull-In of CHA PH5 Disabled
Pull-In of CHB PH1 Disabled
Pull-In of CHB PH2 Disabled
Pull-In of CHB PH3 Disabled
Pull-In of CHB PH4 Disabled
Pull-In of CHB PH5 Disabled

PCIE Frequency : 103

CPU Voltage : 1.325
CPU PLL Voltage : 1.59275
FSBT : 1.37825
DRAM Voltage : 1.9
North Bridge Voltage : 1.4108
South Bridge Voltage 1.5 : 1.55300
South Bridge Voltage 1.1 : 1.20600

CPU GTL Reference 0 : Auto
CPU GTL Reference 1 : -45mv
CPU GTL Reference 2 : Auto
CPU GTL Reference 3 : -45mv
North Bridge GTL Reference : AUTO
DDR2 Channel A REF Voltage : AUTO
DDR2 Channel B REF Voltage : AUTO
North Bridge DDR Reference : AUTO

Load Line Calabration : Enabled
CPU Sread Spectrum : Disabled
PCIE Spread Spectrum : Disabled

thx m8 appreciate this

[In-Fluence]

In my experience/opinion a reboot is less stable than a rounding error.

I'm not sure if a fatal error means its less stable than a rounding error. It sounds worse but who knows for sure.

Sounds like you need more vcore maybe more Vtt and just get it stable, then tweak the GTL's and try and lower vcore/vtt from that point.

This is where I experience trouble finding the GTLs though. If I up Vcore a notch or two, I will get more stability (if not, total stability), but it will still reboot on the majority of occasions. If I give more or less VTT than 1.32 then the system becomes more unstable. On another note, I got back after a weekend away, started the rig up, and with the most stable GTLs I found so far, it was failing withing 5-10 minues of prime (during 1024k test) which it passed all the time 2 days ago and was stable for almost an hour. Something very fishy here imo.