Q6600 Overclocking, BSOD or round error?

[sobol]

Guys if you overclock your cpu too much and provide not enough voltage are you getting bsod while priming or maybe round error ? Knowing that RAM isn't bottleneck.

My G0 can do 3.4 ghz with 1.375v 12h prime stable.
When I try 3.6 ghz the the trouble comes.

I can boot up with 1.375v fine but it can only survive priming one or two lines and then BSOD. With each raise of voltage I get few more calculation lines and BSOD again. I've gave up on 1.60v with 12 lines of calculation and BSOD.

[Knight]

Have you tried changing other voltages besides Vcore?

[Infa]

Dunno, If my orthos fails, its allways BSOD...for some reason, not just freezing, if you get a round error, the margin are very small to right vcore i think, if it bsod's then its too little vcore, and freezing i dunno whats the reason, both together? heh..

well well ...somethings not right anyway if it crashes, but it would be easier to fix it if one could recognize and analyze the symptoms.

Regards.

[[XC] flat-four]

Have you tried changing other voltages besides Vcore?

What are your other voltages at? I was having stability issues on my p5b until I took the nb voltage of auto.

[sobol]

9x400(3600MHZ)
1.5500V bios - Idle 50C, Load 87C, Windows shows 1.544V - 3 Calculations = BSOD
1.5250V bios - Idle 49C, Load 75C, Windows shows 1.512V - 1 Calculations = Freeze
1.5125V bios - Idle 48C, Load 75C, Windows shows 1.504V - 2 Calculations = BSOD
1.5000V bios - Idle 48C, Load 79C, Windows shows 1.488V - 2 Calculations = Freeze
1.4500V bios - Idle 46C, Load 69C, Windows shows 1.440V - 1 Calculations = BSOD
9x386(3474MHZ)
1.3750V bios - Idle 42C, Load 68C, Windows shows 1.360V - 4 Calculations = BSOD
1.4000V bios - Idle 42C, Load 68C, Windows shows 1.384V - 4 Calculations = BSOD

8x435(3480MHZ)
1.3750V bios - Idle 42C, Load 62C, Windows shows 1.360V - 2 Calculations = BSOD
8x430(3440MHZ)
1.3750V bios - Idle 42C, Load 68C, Windows shows 1.360V - 2 Calculations = BSOD
8x425(3400MHZ)
1.3750V bios - Idle 42C, Load 68C, Windows shows 1.360V - 12H Calculations = STABLE


JumperFree Configuration Settings
AI Overclocking: Manual
CPU Ratio Control: Manual
- Ratio CMOS Setting: LOOK AT TOP OF THIS POST
FSB Frequency: LOOK AT TOP OF THIS POST
PCI-E Frequency: 100
DRAM Frequency: 1000
DRAM Timing Control: Manual
CAS# Latency: 5
RAS# to CAS# Delay: 5
RAS# Precharge: 5
RAS# Activate to Precharge: 15
TWR: 10
TRFC: 42
TWTR: 10
TRRD: 10
TRTP: 8
DRAM Static Read Control: Disabled

Transaction Booster: AUTO
Clock Over-Charging Mode: AUTO

CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled

CPU Voltage: LOOK AT TOP OF THIS POST
CPU Voltage Reference: 0.63x
CPU Voltage Damper: Enabled
CPU PLL Voltage: 1.60V
DRAM Voltage: 2.2V
FSB Termination Voltage: 1.3V
North Bridge Voltage: 1.4V
North Bridge Voltage Reference: AUTO
South Bridge Voltage: AUTO

Advance CPU Settings
CPU Ratio Control: Manual
- Ratio CMOS Setting: 9
C1E Suppport: Disabled
Max CPUID Value Limit: Disabled
Vanderpool Technology: Disabled
CPU TM Function: Disabled
Execute Disable Bit: Disabled
PECI: Disabled

[MpG]

Originally, I just got rounding errors. But once I went past 3.4, BSOD all the way.

Presently 8-hrs dual-orthos stable at 3.6 (9x400), Vcore @ 1.4250 (BIOS). But as soon as I try 9x401, I'm lucky to get 10 minutes, no matter what voltages I set everything at. Fwierd, in my books. Short on time these days, but I'm presently trying to verify that my memory isn't the problem.

[monza1412]

I would try other frecuency with the same speed result. I also had problems with 400 x 9 in a e6600 - p5kd combo. Maybe you should try 450 x 8, o just do 401 x 9. Just my 2 cents

[BigRigDriver]

The p5k dlx mite be holding you from reaching desired speeds. I would try another none Intel & see what happens.

[[XC] Synthetickiller]

I have mine @ 3.7ghz, same mobo, but crucial ballistix ddr2 800 (slower than yours) @ 4-4-4-12 timings. I believe, I might be wrong, but my Voltage for the cpu is 1.5175 in bios. Asus AI sweet see it as 1.4720v. Maybe its 1.50 in the bios, not sure. Either way, it was prime stable for 8+ hours.

I have no tried orthos, so I don't know if that'll crash my set up.

I have these set to auto:

TWR: 10
TRFC: 42
TWTR: 10
TRRD: 10
TRTP: 8

CPU Voltage Reference: 0.63x
CPU Voltage Damper: Enabled (I kept this enabled)
CPU PLL Voltage: 1.60V

Everything you set a number for, try it on auto just to see if it'll work. Most people fool around with these, but I've had good luck w/ just leavnig these on auto for 411x9. I cannot get any higher with this bios set up. I'll have to go ahead and change setting like you did. Maybe the auto stuff is the trick. This works for me. If you want, I can post my stability tests on cpuz validation if you want.

Good luck, I hope this info helps a little.

[nrs250]

Quote:

1.5500V bios - Idle 50C, Load 87C, Windows shows 1.544V - 3 Calculations = BSOD
1.5250V bios - Idle 49C, Load 75C, Windows shows 1.512V - 1 Calculations = Freeze

87 C LOAD? Are you serious ? I can't believe that that chip is

1) not throttling
2) at imminent failure if allowed to sit at that temp for too long.

[sobol]

Quote:

1.5500V bios - Idle 50C, Load 87C, Windows shows 1.544V - 3 Calculations = BSOD
1.5250V bios - Idle 49C, Load 75C, Windows shows 1.512V - 1 Calculations = Freeze

87 C LOAD? Are you serious ? I can't believe that that chip is

1) not throttling
2) at imminent failure if allowed to sit at that temp for too long.

1 )Hey, we're trying to figure out how to throttle it.
2 )It was twice I've done it for test, and about 10 mins together.
But I'm not afraid to fool around with high temp.

[CERO]

hmmz isnt the Tjunction suppose to be 100c? If that" theory" is true then it should trottle around 100 and he is under it..... but hey I wont let my cpu run under such heat. And if he dont care about it, thats his own choice, it is in the end his cpu.

gL sobol: maybe try 1.55NBv? Or u should put a fan on the NB and PWN, because heat causing often instability, as a quad core would stress your P5K deluxe allot. And If I am not mistaken, because of heat u need even more vcore then what it actually needed for a curten speed, a fan( 40mm on the NB) maybe laying a 60mm fan ontop of the upper PWM, because that 1 gets extremely hot, the other 1 @ the side ( lookin in the perspective when the mobo is standing vertical in your case),doesnt get hot at all.try to touch the PWM ( upper) , it gets extremely hot during prime.

[sobol]

hmmz isnt the Tjunction suppose to be 100c? If that" theory" is true then it should trottle around 100 and he is under it..... but hey I wont let my cpu run under such heat. And if he dont care about it, thats his own choice, it is in the end his cpu.

gL sobol: maybe try 1.55NBv? Or u should put a fan on the NB and PWN, because heat causing often instability, as a quad core would stress your P5K deluxe allot. And If I am not mistaken, because of heat u need even more vcore then what it actually needed for a curten speed, a fan( 40mm on the NB) maybe laying a 60mm fan ontop of the upper PWM, because that 1 gets extremely hot, the other 1 @ the side ( lookin in the perspective when the mobo is standing vertical in your case),doesnt get hot at all.try to touch the PWM ( upper) , it gets extremely hot during prime.

mck2 thank you for advice on this. I think I will try 1.55v northbridge.
About heat I don't think it is case as I've got 120cm blowing on NB/PWN and during calculations they aren't too hot.


Update: I've tried 1.55v nb and I've got BSOD on 4-th calculation.
I've tried everything set to auto, exempt for mem settings, vcore and I've got BSOD in 5-th calculation.

[bwanaaa]

@sobol

looking at your freezes , they all occur early in your stress run. Yet the temps are in the mid to high 60s.

I am running a G0 and a p5k-deluxe and although i can boot and run windows a 400X9, it fails prime in a few cycles as well. i am presently testing 400x8and even at 1.325 volts, had prime95 failure at 5 hours. i boosted voltage to 1.375 volts and my T went up 10*C
1st run
vCore(CoreTemp) 1.2625
vCore(bios) 1.325 vCore(CPUz,speedfan) 1.296
idle temp 35,35,33,31
load temp 56,56,51,51 to 51,51,48,48 stable to 5 hrs


2nd run
vCore(bios) 1.37 vCore(CPUz,speedfan)1.336 vCore load 1.32
idle 39,39,37,35
load 66,66,61,61 to 65,65,60,60 very stable(many hours)


the interesting thing is that the load temps did not stabilize for about 5 minutes. upon starting prime they quickly peaked to 60 deg but then climbed another 5 deg slowly. my guess is that your 68 degrees is the initial jump and that your temps are continuing to rise. in any event, it's too close to 70 (where throttling occurs) to be useful.

[v0dka]

87C with only 1.52v...

Someone needs to step into the watercooling section

[strange|ife]

try setting your subtimings to auto

and up mch, and ich a bit more.

you may just have a g0 that refuses too do much more than 3.4

my old one wouldnt go past 3.4

[sobol]

I will try today with E6550 and see how far it will go. Got 6 of them so one must be doing at least 3.6ghz. Then I will know if its CPU or mobo.

[bito]

What is the FPO/batch number for this cpu ??

[sobol]

What is the FPO/batch number for this cpu ??

Hi
Its: L720B013 from UK's gladiator.

[sobol]

I'm benching here E6550, stepping SLA9X, FPO L720A626, pack date 06/28/07.

I think it might be a good clocker. For now I'm running small FFT's, 10 calculation until now and no BSOD. Speed is 7x500(3500Mhz) at 1.4500v bios, Rated FSB 2000 Mhz, temp ~60C. My Q6600 would already give up.

I guess I've got crap Q6600.

UPDATE:
After playing more with CPU I'm getting same issue at above speed, BSOD after few calculations.
Tomorrow I'll try Q6600 with MSI P35 Neo and Gigabyte P35.

[sobol]

I was testing the CPU today on MSI P35 Neo-F and Gigabyte P35 DS3P( both with newest available bios) and on all of them I had similar results to my Asus P5K Deluxe.

It was all on fresh Win XP 64 installation.
Each of 9x400 or 8x450 was giving me BSOD on small FFT's but I could successfully finish SuperPI 1.5 with 32 mb.

I'm certain now that the CPU is average clocker.

[sobol]

I've received today my new Q6600, this time its L724A900 and I will try to squeeze 3600 mhz out of it, who knows....

[linflas]

I've received today my new Q6600, this time its L724A900 and I will try to squeeze 3600 mhz out of it, who knows....

My G0 and 680i Striker would not stabalize on the x9 multi, was always close but would not dial in. Found my sweet spot Bios set vCore @ 1.5V (actual under load is closer to 1.39-1.41) Set FSB to 1800 (450) X 8 Multi, bump the NB volts a bit to 1.45 to handle the faster FSB, and SB Volts to 1.6-1.65. Set your memory to 800mhz to start with, attempt to adjust it later if you can stabalize.

[virtualrain]

Every board/processor and OS is likely to manifest different failures in different ways so what triggers an error on my system may appear completely differently on someone else's.

I recently OC'd my new replacement DFI board with my old X3220 quad. I overclocked in Vista x64. I think Vista is better at handling application crashes because I never got blue screens and it handles application crashes a lot more gracefully than XP as well.

I found that if Vcore wasn't high enough I would get rounding errors in Prime in the first hour, but often within the first few minutes.

I found that if GTL settings were off, the whole PRIME application would hang and Vista would notify me as such and allow me to terminate it. The good thing with this is that I could see the hung application in the background so I could easily see how long it ran before it froze. Much better than a BSOD if you aren't around when it happens.

My overclocking style is probably different than most in that I start with low memory clocks (run it 1:1 or less than rated specs) while I max out the CPU and NB voltages and see how high I can push the FSB - tweaking the GTL takes the most time. Then I notch voltages back down if I can... but since I'm water cooling, I usually run at the max clock that the max voltages will support. I use small FFT's to evaluate CPU stability and only when it's PRIME stable for 8+hrs on small FFT's do I then OC the memory and use Blend to test the overall system.

[linflas]

My overclocking style is probably different than most in that I start with low memory clocks (run it 1:1 or less than rated specs) while I max out the CPU and NB voltages and see how high I can push the FSB - tweaking the GTL takes the most time. Then I notch voltages back down if I can... but since I'm water cooling, I usually run at the max clock that the max voltages will support. I use small FFT's to evaluate CPU stability and only when it's PRIME stable for 8+hrs on small FFT's do I then OC the memory and use Blend to test the overall system.

Exactly the point I was trying to make, this is how I do it as well, see what you can do with your cpu and motherboard, and how you can manage temps, deal with the ram later. I run 4 1GB sticks, so it is simple for me due to the difficulties in overclocking 4 dimms. I tend to just run stock mhz, and low voltage beacuse I run WCG 24/7 and higher vDimm will just burn out the ddr2 dimms these days.

Prime95 blend for 12-24 hours has mostly never failed me for a stable clock for 24/7 use. My pc's never get shut down, and are always loaded 100% on all cores, so I may be a different than others in what I consider stable.