• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

TPU's Core i7 Overclocking and Feedback

Glad to see you're enjoying it Paulie. It's a great system to be running, and be honest. Was it so expensive that it was not worth the $$$ like so many people claim it to be?
 
Glad to see you're enjoying it Paulie. It's a great system to be running, and be honest. Was it so expensive that it was not worth the $$$ like so many people claim it to be?

shit, my best investment early this year was my damn i7, I would marry it if possible :laugh:
 
Glad to see you're enjoying it Paulie. It's a great system to be running, and be honest. Was it so expensive that it was not worth the $$$ like so many people claim it to be?

LOL. Even I said it was too expensive a couple of months ago. :o Without a doubt, worth the upgrade. Honestly, it came out to only about $100 more than the Phenom II rig I was running, and that's with 6GB of Ram compared to the 4GB with PII. This rig is just sic, and I'm just getting started.
 
nothing like my good ol crucials for $50 shipped. :)
 
damn kid, amazing how hardware drops soo much heh??

thanks fit :toast:
 
TBH the crucials were this cheap for almost 2 months now! It's just that nobody wanted to test the waters with stuff that looks like bargain ram.
 
i'd sell mine if anyone wants them. i'll get a 6gb kit.
 
cream on this.....

FOLDING HELL!

img0237lh1.jpg

:eek: :eek: :eek: wow nice!
 
Here's my first OCCT run after updating the bios on my DFI DK x58. I know it's not long into the test, but I'm dead tired. Currently running 4.0 at 1.36v. Damn, I love these chips.

17920OCCT.png

So you're getting near 60's on water with 3,9Ghz... Is that near normal or is something holding you back to get lower temps?
Just wanna know what I might be able to expect here by the time I set up a loop.
 
So you're getting near 60's on water with 3,9Ghz... Is that near normal or is something holding you back to get lower temps?
Just wanna know what I might be able to expect here by the time I set up a loop.

LOL. Those temps are actually quite good. These chips run very hot. There are a TON of people on water still creeping into the 70's.
 
LOL. Those temps are actually quite good. These chips run very hot. There are a TON of people on water still creeping into the 70's.

I know they're hot, I've got one :D
 
I know they're hot, I've got one :D

LOL. You just seemed surprised by my temps. I'm guessing the HK will give you a couple of degrees better. I'm also running a rather wimpy rad right now with a couple of Yate Loons. Planning to change that this week.
 
LOL. You just seemed surprised by my temps. I'm guessing the HK will give you a couple of degrees better. I'm also running a rather wimpy rad right now with a couple of Yate Loons. Planning to change that this week.

I'm planning on a 2x120 Feser Xchanger, they seem pretty nice and are just a little cheaper than Thermochills. On 3,9 I can see those temps being normal though. I run higher on 3,3 on air :p
Is this with or without HyperThreading btw?
EDIT: 8 threads, must be HT then. That's pretty decent then, since HT adds 10C to my temps.
 
Since I think it's time to push this baby to its air-limits before I get to move to water, can some of you please fill me in if I thought this method up correctly:

Step 1: find BCLK wall
- Lower CPU multiplier to like 10-12 or something, lower RAM and uncore speeds to as low as possible, then crank up the BCLK step by step until it fails Windows boot.
Question on BCLK cranking in low multipliers: will extra voltage help? or do I only need extra volts when I crank back the multi?

Step 2: combine BCLK with full multi
- Multi back at 20, RAM and uncore low as possible, crank the BCLK up until it fails. You can probably start halfway between the original BCLK and the limit from step 1.
Question here: BCLK will most likely be lower then with multi at low, and I will definately need volts here right?

Step 3: tickle the RAMsies
- Basically, get the CPU multi back down, crank RAM clocks step by step again, add volts when getting failures.

Step 4: combine
- tick down all max settings a nice bit, combine, and tweak until the sun rises again.

What I "know":
- Cranking QPI voltage up to 1.3 can improve stability according to some sources
- CPU voltage should not be set higher than 1.25 unless I've got a good health insurance on the 920 (which I don't). Again, this is according to some sources, I haven't confirmed this.
- RAM voltage should not exceed 1.65 concerning the in-proc memory controller (I don't feel like killing stuff, I just want to clock it up a fair notch)
- Uncore volts shouldn't exceed 1.45 according to some sources

Other stuff I want to know:
- If I shouldn't set CPU voltage above 1.25, why does it show 1.360 on auto settings in CPU-Z? Am I getting into the danger zone or is Auto/CPU-Z messing with me?
- Is there anyone who actually set QPI/Uncore volts and noticed improvement in stability?
- And of course, did I miss something very important in my steps other than "Watch temps!"?

I really wonder if I can kick 4Ghz on air, I did it before but it was unstable as hell (see CPU-Z in sig)
 
you can run 1.5v on air vcore no problem

1.45v is fine for qpi(vtt)

most chips can hit 4.3ghz on air no problem
 
you can run 1.5v on air vcore no problem

1.45v is fine for qpi(vtt)

most chips can hit 4.3ghz on air no problem


Hey Fit. So, on this DFI board the vtt voltage is qpi voltage?:confused: Still trying to understand some of these new i7 settings.

As far as on air with no problem...maybe, but I've heard several stories of temps in the high 80s and low 90's on air with 1.5v. :eek:

Thrackan....1.25v max on vcore. Are you kidding? I think 1.45v is considered a safe max.
 
Since I think it's time to push this baby to its air-limits before I get to move to water, can some of you please fill me in if I thought this method up correctly:

Step 1: find BCLK wall
- Lower CPU multiplier to like 10-12 or something, lower RAM and uncore speeds to as low as possible, then crank up the BCLK step by step until it fails Windows boot.
Question on BCLK cranking in low multipliers: will extra voltage help? or do I only need extra volts when I crank back the multi?

Step 2: combine BCLK with full multi
- Multi back at 20, RAM and uncore low as possible, crank the BCLK up until it fails. You can probably start halfway between the original BCLK and the limit from step 1.
Question here: BCLK will most likely be lower then with multi at low, and I will definately need volts here right?

Step 3: tickle the RAMsies
- Basically, get the CPU multi back down, crank RAM clocks step by step again, add volts when getting failures.

Step 4: combine
- tick down all max settings a nice bit, combine, and tweak until the sun rises again.

What I "know":
- Cranking QPI voltage up to 1.3 can improve stability according to some sources
- CPU voltage should not be set higher than 1.25 unless I've got a good health insurance on the 920 (which I don't). Again, this is according to some sources, I haven't confirmed this.
- RAM voltage should not exceed 1.65 concerning the in-proc memory controller (I don't feel like killing stuff, I just want to clock it up a fair notch)
- Uncore volts shouldn't exceed 1.45 according to some sources

Other stuff I want to know:
- If I shouldn't set CPU voltage above 1.25, why does it show 1.360 on auto settings in CPU-Z? Am I getting into the danger zone or is Auto/CPU-Z messing with me?
- Is there anyone who actually set QPI/Uncore volts and noticed improvement in stability?
- And of course, did I miss something very important in my steps other than "Watch temps!"?

I really wonder if I can kick 4Ghz on air, I did it before but it was unstable as hell (see CPU-Z in sig)

Your testing methodology is not flawed but I would wonder why you're testing the max BCLK. Every multi changes your max bclk. So the next subject is your QPI voltage...

Things to know about QPI voltage:
  • QPI voltage must be within .5v of the DRAM voltage
  • QPI voltage vs CPU voltage- sometimes you have to raise and lower QPI to "feel around" for if it's better closer to the CPU voltage or better further away from the CPU voltage. EXAMPLE: My 4.2ghz operates best at 1.39v CPU and 1.26v QPI with DRAM 1.62v
  • QPI frequency may be a limiting factor in your overclock, but should not be tweaked as a first step. Set this to AUTO or double your DRAM frequency. Lower your DRAM frequency to lower the QPI frequency.

Then you seem to be worried about voltages. I've run my i7 920 @ around 1.4v CPU 1.35v QPI on average for 24/7 since late November. There has been no degradation of my chip, but I can not vouch for how your chip will handle it. I really wish you the best of luck in your OC :D
 
vtt is qpi.... sorta.

on asus boards "qpi/dram" is a combination of several settings. hench why asus has some stability problems.
 
Your testing methodology is not flawed but I would wonder why you're testing the max BCLK. Every multi changes your max bclk. So the next subject is your QPI voltage...

Things to know about QPI voltage:
  • QPI voltage must be within .5v of the DRAM voltage
  • QPI voltage vs CPU voltage- sometimes you have to raise and lower QPI to "feel around" for if it's better closer to the CPU voltage or better further away from the CPU voltage. EXAMPLE: My 4.2ghz operates best at 1.39v CPU and 1.26v QPI with DRAM 1.62v
  • QPI frequency may be a limiting factor in your overclock, but should not be tweaked as a first step. Set this to AUTO or double your DRAM frequency. Lower your DRAM frequency to lower the QPI frequency.

Then you seem to be worried about voltages. I've run my i7 920 @ around 1.4v CPU 1.35v QPI on average for 24/7 since late November. There has been no degradation of my chip, but I can not vouch for how your chip will handle it. I really wish you the best of luck in your OC :D

I did not know that about QPI, thanks. I'll be fondling around for a while now.
You're right about step 1 though.
Guess that step can be for when I fail an OC on BCLK, to see if that clock does work on a lower multi.

Btw, I'm glad my board just says "QPI voltage" when it talks about the QPI voltage :D Makes life easier.
 
Back
Top