[Solved] 1100T OC BSOD with low temps....Help?

System Specs:

CPU: AMD Phenom II X6 1100T Processor, Black Edition 4.1Ghz

CPU Cooler: Cooler Master V8 Nickel Plated Copper Base Aluminum Fins 8 Heatpipes

Case: Cooler Master HAF X Full Tower w/ SuperSpeed USB 3.0 w/ Window w/ Black Interior ATX Case)

HDDs: Seagate Barracuda XT 2 TB 7200RPM SATA 6 Gb/s 64MB Cache 3.5 Inch Internal Bare Drive ST32000641AS
Seagate Barracuda Green 2TB SATA 6Gb/s 64MB Cache 3.5-Inch Internal Bare Drive ST2000DL003

Graphics Card: MSI Radeon HD 6870 1 GB 256-bit GDDR5 PCI Express 2.1 x16 HDCP Ready CrossFireX Support Video Card with Eyefinity R6870 Twin Frozr II

Power Supply: X4 Series 850W Atx Ps

Motherboard: ASUS Sabertooth 990FX - AM3+ - TUF Series

RAM: 2x Corsair XMS3 4GB PC12800 DDR3 1600MHz 4096MB (9-9-9-24-2T) 1.65v

Ambient Room Temp is around 48 degrees F. (House has terrible insulation. haha)

Ok. I tried overclocking last night (First Attempt), I followed the AMD BE guide here on Tom's. Well I kept increasing the multiplier up til I hit 4.1Ghz (My ultimate goal is a stable 365/24/7 4.3Ghz) but I figured I would see how stable my machine is at 4.1 first.

I booted, everything went good.

Checked CPU-Z. Checked Core Temps. Looked good (I think)

Then started Prime 95 Blend test.

I sat and watched the temps, using Core Temps and the AI Suite II. At no point while I watched it, did it go over 34C. I watched for a good 30 min and it seemed to be staying right around 34C.

I left it on overnight and woke up to a restarted comp. My comp BSOD sometime during the Prime 95 test.

I really don't think that it overheated, it seemed to be running really cool. Also, I checked the results.txt from Prime and saw some errors. What could these be from?

So I am asking for help. What should I do now? What are those errors for? Any help is appreciated.

Here is a screen shot of CPU-Z and Core Temps:

Just in case you can't see the screen:

Core Voltage: 1.380 V

Core Speed: 4113.8 MHz

Multiplier: x 20.5

Bus Speed: 200.7 MHz

Here is the log from Prime95:

[Wed Jan 04 21:23:14 2012]
FATAL ERROR: Rounding was 4096, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Wed Jan 04 21:36:51 2012]
Self-test 640K passed!
Self-test 640K passed!
Self-test 640K passed!
Self-test 640K passed!
[Wed Jan 04 21:52:04 2012]
Self-test 8K passed!
Self-test 8K passed!
Self-test 8K passed!
Self-test 8K passed!
[Wed Jan 04 22:37:40 2012]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 1024, expected less than 0.4
Hardware failure detected, consult stress.txt file.

your voltage CPU need more at 4.1ghz and don't let under auto like voltage CPU / RAM and keep NB at 2000 and other setting like Cool n Quite , PCIE spectrum, C1E , turbo from enabled to disabled

Yea. That is the guide I was using. Must not have read it in depth enough. haha.

I have the RAM set to the manual factory settings. I disabled all of those things. IDK if I touched the NB though and I am sure that I have the voltage on auto.

Well 8 more hours at work til I can go home and fix this.

So I will

1. Up the core voltage to between 1.4-1.45

2. Make sure that NB is manual at 2000

3. run memtest

4. try to stress test this again.



Thank you for the help.

good luck

Six PhII cores at 4GHz is a GFLOPin' machine. Doing it at less than 1.4v is really, really good, but not unheard of.

CPUz might be a little misleading on individual core voltage so don't be 100% certain of that 1.38v. I feel a bit more comfortable using the Asus Probe utility for temps and voltage. Actually, I use 3 utilities and triangulate which seems to generally point toward Probe readings

Assuming you disabled Turbo, have qualified memory and left LLC on Auto, I would suggest dropping your CPU multiplier back to 4GHz and taking manual control of your voltages. I would also experiment a bit with different LLC levels (looks like you are at the extreme level, anyway).

Tinker with your NB volts and multiplier. You should be able to run 2400MHz IMC/NB close to stock volts. For each 10% you increase the IMC/NB speed, memory bandwidth is increased 3-4% and latency is reduced 3-4%. 'Enthusiasts' take the NB beyond 3000MHz, but there tends to be a points of diminishing return beyond 25-2600MHz.

Make sure you save a stock profile -- then save your OC profiles, too.



edit __ it's your HT link speed at 2000MHz (no OC'ing)

Ok. well read up on LLC and it doesn't sound like there is a definitive answer for whether or not to use it.

http://www.overclock.net/t/556097/cpu-llc

It sounds like:

If i raise my Core voltage to anything above 1.38, then LLC may cause surge spikes over the 1.45 limit....in turn damaging my CPU.

Maybe I am misunderstanding this? what levels can you set LLC to? (other than auto and off)


EDIT:


SO,

Let's say I up my core voltage to 1.42 and my NB to 2500. What would be an appropriate LLC level to maintain stability and not risk excessively high spikes?

Ok, thanks.

Yea, I think I will leave LLC on auto. I read through that section and still don't fully understand it.

This confuses me. This means that under load, my Core voltage and multiplier will increase from what I already set it at correct?

Saying I set my core voltage to 1.42 wouldn't this be dangerous even if I picked the lowest 25%? That would push it over the factory range limit of 1.45v.

(I did disable turbo. I disabled all of the "features" according to the guide)

Im sorry. I am just having trouble catching on to the whole LLC, VDroop, offset concept.