TPU's Core i7 Overclocking and Feedback

[brandonwh64]

Both of you are being wusses about voltage. Even D0s on average need 1.35-1.4 to get stable 4.0GHz. Especially with multi-GPUs. Fits and I had a discussion today about that.

but i pass stability tests great.

im still getting BSODs without overclock

 

[Binge]

but i pass stability tests great.

im still getting BSODs without overclock

OS/PSU/Symptoms other than BSOD. There has to be a precursor to the BSOD.

 

[brandonwh64]

from what i have gathered on my types of BSODs i get that its either a OS problem or a driver problem.

0x00000A
0x0000D1
0x00001E

These are the three i have been getting alot

 

[Binge]

from what i have gathered on my types of BSODs i get that its either a OS problem or a driver problem.

0x00000A
0x0000D1
0x00001E

These are the three i have been getting alot

Interesting fact, your CPU can cause OS and driver problems with instability. BSOD dumps are extremely vague/prone to error since the computer itself is basically f*cking up.

 

[mastrdrver]

Maybe your raid is causing the problem (driver related)?

I havn't ran a raid since I first did it on my K6. It was quicker, but when I lost it I realized it just wasn't worth it to me.

 

[{JNT}Raptor]

Ok just got another BSOD 0x0000001E ntoskrnl.exe.

googled this one and it says that there is a bug in windows 7 x64 ultimate that plagues ntoskrnl.exe and causes it to BSOD.

Hmmm....I've been on Win7x64 Ultimate....like most....well before the Official release.....and I've never gotten any BSOD related to "ntoskrnl.exe" .....no BSODS at all actually...unless you count the BSODS while finding the limit of the overclock....even then I re-imaged over my install for a fresh environment after.....Win7 has been SOLID for me from the start.

It almost sounds like your overclock wasn't 100% and corrupted something that had a degrading effect over time.....either that or you may have a hardware issue you haven't found yet.

Hope something helps.

 

[t_ski]

So great I got a technet subscription.

That's what I did:

Hmmmm... Should I pay $200 for one OS license, or should I pay $265 and get 10 OS licences and 10 Office licences....

 

[brandonwh64]

Ok i raised the PLL to 1.9V and the QPI to 1.4V and i did 2 hour OCCT CPU test on max level and a intel burntest on max for 5 passes. NO LOCK UP

I will let it idle to see if it BSODs

 

[t_ski]

Try PLL at 1.8v. I have not heard of anyone ever needing to raise it, especially only for 4.0GHz.

 

[brandonwh64]

1.8 is stock tho?

 

[PaulieG]

Both of you are being wusses about voltage. Even D0s on average need 1.35-1.4 to get stable 4.0GHz. Especially with multi-GPUs. Fits and I had a discussion today about that.

This is particularly true with later batches/pack dates, at least in my experience. For me, it was just about chronological. My first 920 was my best, and they slowly got worse. After my 5th or 6th chip, they all needed at least 1.325v to a stable 4.0ghz. My first few all did 4.0 on less than 1.30v

Know that I am not referring to particular batches, as the quality of chips seemed not to parallel particular batches, only pack dates were consistent.

The problem that people are running into with higher vcore chips is the heat that comes along with it. There are only a couple of air coolers that can handle 1.375v on an i7.

 

[brandonwh64]

mine was mid 2009 batch

 

[Chicken Patty]

This is particularly true with later batches/pack dates, at least in my experience. For me, it was just about chronological. My first 920 was my best, and they slowly got worse. After my 5th or 6th chip, they all needed at least 1.325v to a stable 4.0ghz. My first few all did 4.0 on less than 1.30v

Know that I am not referring to particular batches, as the quality of chips seemed not to parallel particular batches, only pack dates were consistent.

The problem that people are running into with higher vcore chips is the heat that comes along with it. There are only a couple of air coolers that can handle 1.375v on an i7.

ahem p

 

[Chicken Patty]

Guys, about voltages and BSOD's, this guide helped me identify the cause of the instability at times. For those who haven't checked it out, please do so.

http://www.evga.com/forumsarchive/tm.asp?m=100494809&mpage=1&key

 

[mastrdrver]

Ok i raised the PLL to 1.9V and the QPI to 1.4V and i did 2 hour OCCT CPU test on max level and a intel burntest on max for 5 passes. NO LOCK UP

I will let it idle to see if it BSODs

http://img.techpowerup.org/100303/stresstestagain.jpg

Have you tried raising your IOH voltage?
Do you have all power saving stuff turned off in the Bios?

With IBT and LinX, you need to run it min 25 pass to see inconsistency. When you have everything running at minimum voltages, your GFlops from one run to the next should hardly vary. Make sure you run it with nothing else running, including RT or any taskbar stuff. I was able to consistently get the variance down to no more than .5 GFlops from highest to lowest number. One or two outside this is not uncommon on a full install but no more.

Then again, that is only true if you have not borked your OS install. Also, it helps some i7s to keep VVT with in .3-.25v of vCore. My i7 is like this and it needs to be with in that window under load or it is guaranteed to fail. See link in sig for details.

 

[Binge]

Bwahahaha!

 

[mastrdrver]

Yep that's what I found out, no need for super high Ram speeds (and therefore uncore) to achieve fast performance on 1366. Though, I hear its different on 1156.

I can run 1600 CAS7 but get same performance on 1200 CAS5 (when using 200 bclk).

 

[t_ski]

1.8 is stock tho?

Yes, it is. I'm at 4.4GHz and mine is still at stock.

Guys, about voltages and BSOD's, this guide helped me identify the cause of the instability at times. For those who haven't checked it out, please do so.

http://www.evga.com/forumsarchive/tm.asp?m=100494809&mpage=1&key

That's at least the third time that thread has been linked/listed in this thread.

 

[Chicken Patty]

Yes, it is. I'm at 4.4GHz and mine is still at stock.



That's at least the third time that thread has been linked/listed in this thread.

Yep, but I don't expect anybody to look through the whole thread.

 

[t_ski]

Just thought it was funny

 

[Chicken Patty]

Just thought it was funny

Yea New people are coming into this thread all the time, so why not.

 

[t_ski]

 

[Binge]

Yep that's what I found out, no need for super high Ram speeds (and therefore uncore) to achieve fast performance on 1366. Though, I hear its different on 1156.

I can run 1600 CAS7 but get same performance on 1200 CAS5 (when using 200 bclk).

That screenshot has nothing to do with slow vs fast memory. In fact it's 2k ram vs 2k ram.

 

[brandonwh64]

Ok i will back down my pll to 1.8

its just since ive done some minor voltage bumps to certain things, it has lowered the BSODs

I did 15 passes of intelburntest last night and no BSOD. plus after it was done it idled all night and no restart

 

[Binge]

This picture should be more descriptive of the kind of results this ram is seeing.