DAMNED QX9650!! What's wrong?? Mistery

[kito]

Hallo everyone,
I'm trying to setup a new system,
here are my specs:

-MOTHERBOARD: ASUS STRIKER EXTREME II 790 ULTRA SLI BIOS:1401
-CPU: QX9650
-VIDEOCARD: NVIDIA 9800GX2
-RAM: ddr3(corsair dominator PC3-12800 CL7)
-PSU: Enermax Galaxy 1000W
-COOLING:Phase


There is nothing I can do to overclock my CPU the way I want, I cannot get it stable over stock 3,00 Ghz.. and this seems really strange to me, since I'm under phase, and lots of people managed to get this chip 100% stable with same board over 4Ghz by air.

I was able to boot the system by setting the right Vcore all the way up till 5Ghz.. but i can't get it priming for more than a few minutes.

Over the 3.0 Ghz this is what happens when priming all 4 cores (Small FFT):

-Vcore to low: Rounding errors soon or later on all cores.
SLOWLY rising the vcore..
Rising the Vcore increases stability but as soon as all cores seem not to give any rounding errors in the first few mins, system suddenly resets or locksup when priming. What the hell?

Cpu temp is -20 sub zero (under load), it is not overheating.

All this happens if i raise the FSB and keep the 9x multiplier but also if i keep the FSB still and try to raise the multi.
As I said I can almost always boot into windows but i can never prime for more then 5 mins when overckocking.
CPU primes with no errors at all when @stock speed and voltage.


It's not my first overclock i've been using phase for several years and always achieved good overclocks...

By the way:
Memtest 100% OK.
Memory is working. Anyway I'm trying to get it stable using only a stick of ram.
Timings are not aggressive and voltage is OK.

Question:
Do you think it's possible that I might have used too much compund on the cpu?

 

[glock19owner]

I am not familiar with your board at all...but here are the settings I run at 4.275 Ghz daily running on my QX...

Vcore = 1.44
VTT (FSB Term. Voltage) = 1.325
PLL = 1.677
Memory = 2.1v (depends on how your run your divider) (1800Mhz is what I am running)

GTL 1 = +40
GTL 2 = +50
GTL 3 = +50
GTL 4 = +40

Also might have to play around with your NB voltage...

 

[orion456]

Northbridge and FSB voltage needs to be quite high for those chips to OC.

 

[kito]

Thanks for replying..

I tried to raise Nb sb PLL Vtt everyting... doesn't seem to be the real problem..
The fact i can't overclock it from 3.0Ghz to 3.16Ghz (this should be possibile keeping stock vcore) without PC resetting under load makes me suspect there might be an issue with my PSU..

Today I will borrow a psu to see if my problem is alimentation related..

 

[glock19owner]

What memory timings are you trying at and at what divider?

 

[orion456]

Thanks for replying..

I tried to raise Nb sb PLL Vtt everyting... doesn't seem to be the real problem..
The fact i can't overclock it from 3.0Ghz to 3.16Ghz (this should be possibile keeping stock vcore) without PC resetting under load makes me suspect there might be an issue with my PSU..

Today I will borrow a psu to see if my problem is alimentation related..

I would suspect the memory then. Take out one stick, and see if you can OC then. If not, swap them. Make the timing very loose. Make sure you have the appropriate voltage for the memory.

 

[kito]

What memory timings are you trying at and at what divider?

(It was not the PSU)

My ram is meant to work @1.90V @1600 7-7-7-20

But I tried every possible combination
Linked @ 1333:1333
lowering ram to 1066mhz

I have tried Corsair ram and OCZ ram, also using one stick and loosing the timings and frequency did not help.

OK, I have news..

Now system resets also at stock speed when priming (using one stick of ram only, memory did pass memtest without any errors)..

 

[kito]

I will try to repeat the test using stock cooler instead of the phase.. maybe there is something wrong with the insulation and water condensation could be causing all this.
Do you think bad insulation could be causing this behaviour under load?

 

[kito]

I will try to repeat the test using stock cooler instead of the phase.. maybe there is something wrong with the insulation and water condensation could be causing all this.
Do you think bad insulation could be causing this behaviour under load?

Im using stock cooler..
I coulnt boot @3ghz, I loaded failsafe settings (1.90Ghz)
CPU IS TEMPERATURE IS 73° IN BIOS @1.9Ghz..

Something must be overheating..

I decided to SWAP the qx9650 with my old E6400(which has always worked perfectly)..

Running @stock speed 2.13Ghz

Priming for 2 minutes then SUDDEN POWER OFF!

Before priming CPU temperature was 42°C
When priming, in a few secs temperature raise to 100C° and after a few PC Shut down.

 

[orion456]

Starting to look like the mobo, since I doubt the CPU is bad. Any chance you could switch mobos?

You might also try disconnecting everything and using a new HD to boot. I had a system where the HD was causing resets, took forever to find that.

 

[kito]

Starting to look like the mobo, since I doubt the CPU is bad. Any chance you could switch mobos?

You might also try disconnecting everything and using a new HD to boot. I had a system where the HD was causing resets, took forever to find that.

Tnx for reply...

I'm sorry i did post something wrong..

Temps were that high using stock cooler because i did some mess with the fan..

I MAYBE have found the probem..

I switched the QX9650 with my old e6400 and my older Pentium D840XE both of them priming ok...

Using hardware monitor i noticed that the "DDR3 Termination voltage", which has always been set to AUTO was at about 0.82V(auto) using E6400 and on D840XE.

As soon as i switched to the QX9650 again i noticed the DDR3 Term.Voltage (always on AUTO) had reached 0.99V...

I set it to Default DDR3_ref and guess what???

EVERTYTHING IS WORKING...Im running and priming 3.20Ghz on air without resetting or anything...

This really makes me happy..!

What i would like to ask you is:

If DDR3 Termination Voltage was set to AUTO all the time:
- Why did AUTO mean 0.82V using E6400 and 840XE and why 0.99 when using QX?? (always using only 1 stick of ram)

Case seems to be closed! The only victim of this issue is the post Led, which I stepped on accidentally

 

[Airbornederekc]

Wow isnt that just something you dont see everyday.

 

[orion456]

If DDR3 Termination Voltage was set to AUTO all the time:
- Why did AUTO mean 0.82V using E6400 and 840XE and why 0.99 when using QX?? (always using only 1 stick of ram)

Case seems to be closed! The only victim of this issue is the post Led, which I stepped on accidentally

So it was the memory, interesting . . . sometimes auto is set from a table of values and it may be an error in the table for the QX9650. I've seen several table errors on ASUS boards.

 

[El<(')>Maxi]

Make sure you have good contact (test mount) when you re-mount the phase unit.