I7 and BSOD

[Daddyjaxx]

Is it just me, but when my OC testing fails on the I7 it throws more BSOD's than my other CPU's. I'd hardly ever used to get a BSOD. Prime would just find an error and stop. Now the failures are always a BSOD.

My CPU does seem to like 21x more than 20x. I'm trying 4150 now with less volts than I was getting at 4100 with the 20x. Only in the low to mid 60's so far. I think this puppy has some more life in her considering I'm only at 1.26.

 

[Category 5]

Same thing here. I am noticing that core failures seem to mean something with cpu (re. volts) and BSODs mean tweak the ram or qpi link. Seems the memory controller in these things is very sensitive and I myself am having a hard time getting as stable as I'd like.

Bummer...I had P95 running for 33 hours when i got home the other day. I was happily preparing to declare stability when, as I turned my back for one second I heard that dreaded beep. Restart! I was praying for a "Windows has installed updates" message, but alas it was a "A serious error occurred" --

Back to the drawing board. I have a feeling a lot of these "stable" overclocks aren't really "military grade 24/7 stable"

 

[Daddyjaxx]

You're kidding...right? 33 hours of P95 is good enough unless you plan on folding or doing something with 24/7 100% load.

 

[Category 5]

You're kidding...right? 33 hours of P95 is good enough unless you plan on folding or doing something with 24/7 100% load.

Normally 24 hours = stable, but I was out. But a BSOD happens for a reason so....

 

[Category 5]

You're kidding...right? 33 hours of P95 is good enough unless you plan on folding or doing something with 24/7 100% load.

Well damn if I can't figure this thing out. D0 only at 3.6, ram underclocked to 1440 (tested 24hrs at 1600 with stock cpu), cpu at 1.21, ram at 1.55 qpi at 1.3625

OCCT stable for 10 hours, OCCT linpack stable for 24hrs, p95 small, large and blend stable for 24hrs

browsing the web and BSOD! What gives with this i7 platform and the BOSD? I never experienced so many in my life. It's driving me batty! 3.6 isn't even that crazy for a d0 920.

 

[Daddyjaxx]

There are more to BSOD's than CPUs and Memory...hard drive, GPU, Windows throwing up.

 

[baditude_df]

You're kidding...right? 33 hours of P95 is good enough unless you plan on folding or doing something with 24/7 100% load.

If your CPU can do Small ffts or LinX for 24hours then it should be able to fold without errors. Any more time than that and you're just wasting electricity IMO. Obviously there are differences in the way the programs work the CPU, but stress testing programs just flat out work a CPU harder than folding does. That's clear in the temperature output observed while doing either or. You can say you need more time, but what is the yardstick there? I mean, how do you really know?

Well damn if I can't figure this thing out. D0 only at 3.6, ram underclocked to 1440 (tested 24hrs at 1600 with stock cpu), cpu at 1.21, ram at 1.55 qpi at 1.3625

OCCT stable for 10 hours, OCCT linpack stable for 24hrs, p95 small, large and blend stable for 245hrs

browsing the web and BSOD! What gives with this i7 platform and the BOSD? I never experienced so many in my life. It's driving me batty! 3.6 isn't even that crazy for a d0 920.

I can see that you're trying to get away with low voltage on the CPU and RAM, but in my experience, that's where the BSOD's come from. Bump them both up to take them out of the equation if you haven't already done so.

And

large and blend stable for 245hrs

Holy cow! That's taking things to a level I never knew existed.

 

[Category 5]

And
Holy cow! That's taking things to a level I never knew existed.

lol you know that's a typo right!? I meant 24hrs

So was ram volts. It is at 1.66. I got one of those dumb split MS keyboards and can not type on it!

You think CPU volts is still too low? Even for 3.6? That seems mild compared to what I'm seeing on other D0 chips

 

[baditude_df]

For 3.6GHz it's not low. But barring everything else, you could have a chip that just doesn't like to clock. Not a definite, but a possibility. Have you done a google search on your batch#? It might give you some insight.

 

[GMoney86]

I was having a bunch of blue screens, too, but I found the solution. First, it is important to note the type. Some blue screens are related to the processor (and I have seen them when I didn't increase the voltage enough), but some are related to other hardware or even software. As I said, some BSODs were from my processor, like one saying that there was a critical hardware error and another talking about a missed interrupt, but most were being caused by corrupt windows files. I reinstalled windows 7 and since then I have had only one BSOD, which is when I lowered the voltages to stock levels on a 4 GHz overclock. Your BSODs might just be due to corrupt files on the hard drive.

 

[Daddyjaxx]

It's a crapshoot. Sometimes I can do hours of Linx at 1.2375 at 4.0 and it doesn't fail. Other times it BSOD's. I upped it one notch and it never failed. I know I've got a pretty good cpu, but my P6T is holding it back. If I didn't have the throttling bug, I know I could eek 4.4 out of it. Not for day-day, but just to see how high it could go.

 

[baditude_df]

HT, no HT?

 

[Amtrak]

If I didn't have the throttling bug, I know I could eek 4.4 out of it. Not for day-day, but just to see how high it could go.

It's not a bug, per se. It's just an additional multi throttling which many other mobos allow you to adjust or turn off entirely through the BIOS. The Rampage II Extreme is supposedly like this, and ASUS is reserving that feature for it's "premium" boards. Since you have the V1, if you're serious enough about it, you can flash to a special BIOS (version 0007 or 0006). V2 users can also do that, but must first flash to a V1 BIOS, IIRC.

 

[Category 5]

HT on. What's this throttling bug?

 

[Amtrak]

HT on. What's this throttling bug?

If you set the 21x multiplier in the BIOS, the actual multi can throttle down under high Vcore (around 1.4, according to the XS thread) or high heat. This is because of the way the 21x multi is implemented, as it is essentially forcing the regular turbo boost always on.

For me, the turbo throttling kicks in very slightly when any CPU core hits about 84-85C. These are not harmful temps, but it throttles anyway and can cause instability. Various other boards allow this to be adjusted or disabled in the BIOS, from what I understand.

 

[Category 5]

If you set the 21x multiplier in the BIOS, the actual multi can throttle down under high Vcore (around 1.4, according to the XS thread) or high heat. This is because of the way the 21x multi is implemented, as it is essentially forcing the regular turbo boost always on.

For me, the turbo throttling kicks in very slightly when any CPU core hits about 84-85C. These are not harmful temps, but it throttles anyway and can cause instability. Various other boards allow this to be adjusted or disabled in the BIOS, from what I understand.

Ah. I have turbo disabled so I'm at 20x, and I allow speedstep to lower under light load.

Do you think QPI voltage is supect here? I can bump up the cpu but was getting 3.33 (with trbo on so 3.5) with 1.13volts, and now 3.6 (turbo off) is at 1.2065 and I may need more?

 

[Amtrak]

Did you test for stability using the turbo at 3.5?

 

[Category 5]

Did you test for stability using the turbo at 3.5?

Yeah. 6 hours occt, 100 passes linx and 24 hours small fft. Ram was at 1333 I think.

All tests passed.

With the current setting did 33 hours of small fft, 24 hours blend, 6 hours occt, 100 passes of linx and 24 hours occt linpack. When i stopped the occt linpack (which had no errors) i got a message "the program has encountered an error and has to be terminated".

I restored the OS from a known good ghosted copy and tried again today. Got the BSOD while installing nvidia drivers, and some other odd behavior (programs feeling sticky, etc).

Other change was from OCZ Powerstream 520 to a Corsair TX850 - pretty sure that swap didn't introduce instability!

 

[Amtrak]

Why did you restore the OS for a program error? Sounds to me like that may have caused a lot more problems.

 

[Category 5]

Why did you restore the OS for a program error? Sounds to me like that may have caused a lot more problems.

Didn't restore for that. Restored because while i was experimenting I had a lot of crashes and BSODs and needed to rule out the possibilty of a corrupt OS with these new problems I'm having.

...little things like runtime c++ errors in IE, and strange overall program behavior but not flat out crashes