QX9650 OC Stability Problems

[mph]

Hello all, completely new to overclocking and looking for some advice.

I've read the stickied OC guide by graysky but currently not sure what to do.

My goal is 4.0 on air. I am in windows, but the prime95 test will BSOD my computer.

 

Note:

CoreTemp vs RealTemp - I'm not sure which you prefer, but RealTemp = CoreTemp + 5

E.g. CoreTemp displays 40, RealTemp will show 45.

I've only included CoreTemps, but I used both to monitor stress testing.

 

Computer

Vista 64 bit SP1

CPU: QX9650 (currently unstable at 400 FSB x 10 Multiplier)

Cooling: TRUE 120

MoBo: evga 750i FTW

RAM: 2 x 2gb corsair 1066/8500

GPU: GTX 280

 

Background

1. 333 x 10

Originally @ stock settings (333 x 9), changed to 333 x 10.

Ran Prime95 overnight, and stopped test after 8 hours (stable).

Did not change voltages from stock:

CPU Core: 1.25 (may have been higher, the setting was auto)

CPU FSB: 1.1

Memory: 1.9

nForce SPP: 1.3

NF200VL: 1.2

 

2. 350 x 10

Pushed onward to 3.5

Unlinked memory so only change is to CPU.

CPU Core 1.3125 (from auto)

CPU FSB: 1.1 (from auto)

Memory: 1.9 (from auto)

nForce SPP: 1.3 (from auto)

NF200VL: 1.2 (from auto)

 

GTLVREF Lane 0 - 0 (from Auto)

GTLVREF Lane 1 - +55 (from Auto)

GTLVREF Lane 2 - 0 (from Auto)

GTLVREF Lane 3 - 0 (from Auto)

 

Note that these voltages (aside from CPU Core) were unchanged from stock settings

I'm not going to tinker with GTL yet because I'm not sure what it does.

 

Ended torture test after 1 hour (stable)

 

3. 375 x 10

First signs of potential stability issues.

Voltages unchanged (Core - 1.3125).

Torture test 1 - computer restart after 1hr10m

Torture test 2 - computer restart after 1hr55m

 

Thoughts: Stable enough to proceed to target of 4.0 (not sure if this was a good assumption)

 

4. 400 x 10

4A. Voltage unchanged (Core - 1.3125)

Post, but BSOD @ Windows.

 

4B. CPU Core 1.325 (from 1.3125)

Post, but BSOD @ Windows.

 

4C. CPU Core 1.3375 (from 1.325)

Post, but BSOD @ Windows.

 

4D. CPU Core 1.35 (from 1.3375)

Post, but BSOD @ Windows.

 

Thoughts: at this stage, I thought perhaps it wasn't the Core voltage that needed modifying and tried something else

 

4E. CPU Core 1.35 (unchanged from 4D)

CPU FSB 1.15 (from 1.1)

 

Get into windows ok!

Temps at idle according to CoreTemp are:

Core 0 - 39

Core 1 - 35

Core 2 - 28

Core 3 - 31

 

Note - is it normal to have such a large range of temps? Core 0 is usually 10 degrees warmer than Core 2!

 

Stress test results in immediate BSOD

 

4F,G,H ommitted due to pointless testing (reduced voltages instead of increasing)

 

4I. Decided to try auto voltage for the other (non-core) settings to see if this is a better judge of voltages than I

CPU Core: 1.35 (unchanged from 4D)

CPU FSB Auto - 1.3

Memory Auto - 1.9

nForce SPP Auto - 1.4

NF200VL Auto - 1.2

 

Get into Windows Ok!

Temps at idle according to CoreTemp are:

Core 0 - 40

Core 1 - 37

Core 2 - 30

Core 3 - 33

 

Stress test results in immediate BSOD.

 

4J. Only changed CPU Core to 1.35625 (from 1.35)

Windows is ok, but testing again results in BSOD.

 

4K. Cloned 4I, but changed all GTLVREFs to Auto (from manually input stock settings)

BSOD @ windows

 

4L. Current settings!

 

CPU Core: 1.3625

CPU FSB: 1.3 (auto)

Memory: 1.9 (auto)

nForce SPP: 1.4 (auto)

NF200VL: 1.2 (auto)

 

GTLVREF Lane 0: 0 (auto)

GTLVREF Lane 1: +65 (auto)

GTLVREF Lane 2: 0 (auto)

GTLVREF Lane 3: 0 (auto)

 

Stress testing results in immediate BSOD. Stability while IDLE is tenuous at best.

 

I'm not sure what to do now.

The QX9650 is only rated to go up to 1.3625, and I don't want to turn my CPU into a pile of slag.

 

What should I do?

Am I doing anything wrong?

What could be causing the BSOD when I stress test?

Any other hints/tips?

Sorry about the length, I just thought the more details, the better.

I appreciate any help you can offer :thumbs-up:

 

TL;DR - BSOD when I stress test @ 400 x 10. Voltages (or other variable) may need tweaking. Halp!

 

Edit: I can provide photos of my BIOS if required. I haven't tweaked with any of the other settings, so they should be at stock.

Edited June 28, 2009 by mph

[MJCRO]

- Try to drop the multiplier down to 9 or 8 and give that a shot first

- Your motherboard may also be a factor causing OC instability

- I'm not sure on your system, but make sure that temps are holding ~55 C at peak load

- Really search through the net for people with similar specs as you and their OC's for comparison.

- Also your CPU stepping may be a factor, if it is C0, chips like that tend to run hotter than the newer stepping E0 which OC best.

Edited June 28, 2009 by MJCRO

[mph]

@MJCRO - Thanks for the quick reply

 

1. FSB/Multiplier tweaking

Earlier OC settings with lower FSB (333, 350, 375 to some extent) and a multiplier of 10 appeared to be stable.

My OC target is 4.0 - i'm not fussed whether I achieve that with a multiplier of 8, 9, or 10 (although 10 seems ideal to me).

I'll try fiddling with the FSB/multiplier until I can achieve stability.

 

2. MoBo

I'm not sure what I can do to check if it is my MoBo causing instability.

Suggestions?

 

3. Temps under earlier OC settings (3.333, 3.5, 3.75) while stress testing with Prime95 hit a maximum of 53 degrees in Core 0.

The other 3 cores were at various temperatures below this.

 

4. Am currently searching for similar situations. Yet to find anything concrete that may help.

 

5. CPU stepping?

 

Question - At this stage, I should be increasing voltages for stability purposes correct?

If my CPU has a max voltage of 1.3625 (as rated by Intel), is it safe pushing it beyond this?

I am already sitting at 1.3625 and am cautious of any damage that I might do to my CPU.

 

Please also find attached, a screenshot of the DEFAULT and CURRENT system voltages.

Edited June 28, 2009 by mph

[MJCRO]

I guess yours temps aren't causing BSOD's so, its got to be the multiplier, tha'ts my best guess, try lowering it to 9, and pushing it 400~ voltages set to 1.275~, I'm pretty confident that you'll be able to hit 4.0 Ghz no problem.

 

*Yes it is safe to raise voltages to 1.3625 and above but not above 1.45 for that's no so good for 45nm chips and will decrease the chips overall life.

 

Also in the BIOS, disable any EIST's controls that prevent the CPU from OC'd by the motherboard itself and set all values to standard if possible for best possible OC.

 

http://www.intel.com/cd/channel/reseller/a.../eng/203838.htm

 

Also forgot to say welcome to OCC :thumbs-up:

 

Get back to me if your still having problems with your OC.

 

EDIT:

 

CPU Stepping is a revision on the CPU itself and updates made to the chip to allow to perform better.

 

Most older Q/X9xxx series are C0 stepping, but if your system was recently built, you most likely will have E0

 

Get this App to get indepth info. on your PC:

 

http://www.cpuid.com/cpuz.php

Edited June 28, 2009 by MJCRO

[mph]

Oh dear

I reduced the multiplier to 400 x 9 (3.6).

Overnight, I left my comp on for stress testing.

Things looked ok when I woke up.

Temps peaked at 58 (realtemp), and after 8 hours of stress test there were no errors reported by Prime95.

 

Goto work, fairly happy with a 20% OC (especially considering it's my first attempt).

Come back, turn comp on -> MoBo begins playing up.

My comp POSTs ok, but the MoBo FSB displays on start up to only be 1066 (266 x 4), instead of 1600 (400 x 4) as per my settings in the BIOS.

The computer also seems very sluggish while loading up.

 

I've now reset everything to default settings.

The MoBo now displays correctly (1333 @ stock), and everything seems ok (aside from my lost work).

What's wrong with my MoBo?

Have I majorly stuffed something up?

I'll try and get a photo of the start up to show you what I mean.

 

P.S - thanks for the kind welcome

 

Edit: Attached photo of start up.

My settings in BIOS are 400 x 9, but as shown in the picture, it only displays as 1066 for some reason...

Edited June 29, 2009 by mph

[MJCRO]

Well if your really stressing about temps. which I don't blame you then, save some money and get this

 

http://www.newegg.com/Product/Product.aspx...N82E16835227005

 

Its a all-in-one cooling system for CPU's, when I get enough money I'm definitely picking one up.

 

I'm pretty sure if your temps came down, you could definitely hit 4.0 Ghz on a 9 multiplier.

 

But sometimes not all chips can OC far like some do, just luck of the draw.

 

I doubt anything is to do with your 750i, because I've seem people hit 4.5 on water with that.

 

The boot up with 800/1066 is weird, but is that after a BSOD or something, and check in windows with CPU-Z (I left a link before) to see current speeds.