3570k Overvlock

[rhyseyness]

I've tried adjusting the vdroop offset manually. Tried both 12.5% and 100% (the highest and lowest options) and neither seemed to help the situation, dropped the vcore when I tried both and still BSoD under load :-/

[Guest Gazzer96]

90oC fro a CPU? My Laptop with an Intel i7 Q 740 only reach 80oC when gaming!

[Wats]

Digital compensation level sounds like LLC. Try raising that to about 75% or right under the highest setting.

Lower your CPU PLL to 1.7V

Set Dram to 1.52V.

You might be able to disable PLL overvoltage, leave it as is for now.

Boot up and run that through a small stress test. If I'm somewhat correct you're temps will be very warm @ 1.36V

Make sure to watch your Vcore with CPU-Z and temps while stressing

[rhyseyness]

Digital compensation level sounds like LLC. Try raising that to about 75% or right under the highest setting.

Lower your CPU PLL to 1.7V

Set Dram to 1.52V.

You might be able to disable PLL overvoltage, leave it as is for now.

Boot up and run that through a small stress test. If I'm somewhat correct you're temps will be very warm @ 1.36V

Make sure to watch your Vcore with CPU-Z and temps while stressing

Thanks, I'll try that now. I think the digital compensation is auto or high, so I'll try high. Also CPU LLC (Assuming that's the same as the vdroop control) is a percentage, so I'll work out the percentage for that to be at ~1.7V and report back. I can disable PLL overvoltage, but I'll leave it as you suggest. I'll change DRAM voltage to that too. Wish me luck... If I'm stable with those changes at 4.5GHz, I'll drop the Vcore bit by bit, and see what happens. Thanks! I'll report back ASAP :-)

[Wats]

Woah hang on!

LLC and PLL are quite different!

LLC controls Vdroop, Set that to 75% or right under the highest setting so that your Vcore stays near 1.36V while under load.

PLL extends OC range. Lower that to 1.7V because this is just a small OC.

Just clarifying. Don't burn the house down!

[rhyseyness]

Woah hang on!

LLC and PLL are quite different!

LLC controls Vdroop, Set that to 75% or right under the highest setting so that your Vcore stays near 1.36V while under load.

PLL extends OC range. Lower that to 1.7V because this is just a small OC.

Just clarifying. Don't burn the house down!

Haha, Thank you for the clarification, I don't actually know any of this stuff bar what I've got from the limited research which I've done, and this is my first time.

Digital compensation was either auto or high, so I changed it to high. This didn't affect stability or voltage of the CPU under load as far as I could see. When I then dropped the vcore, I got BSoD. Setting the memory voltage to 1.52 caused BSoD too, so I put that back to auto. I don't see any option to adjust the PLL voltage, only to enable or disable overvoltage. Gonna now try vdroop offset control (This is LLC?) to 87.5% and see the results. I'll leave the digital compensation level at high for now, as it didn't appear to affect stability

[rhyseyness]

Ok so, at 87.5% LLC, on a small stress test, stability is good, and voltage stays at 1.328, occasionally jumping to 1.32.

Temperatures are naturally still high.

Going to now attempt dropping vcore again and see if stable. Possibly dropping the vcore and increasing LLC to 100%? Is this a good idea?

[rhyseyness]

Not stable at 87.5% LLC at 1.34V, nor at 100% LLC at 1.34V

Gonna get to bed now, as it's midnight here.

I'll check back in the morning.

Hopefully you'll have a brainwave Wats ;-)

Cheers

[Wats]

Lower Vcore normally doesn't need more LLC. 75% is a great amount. 100% can cause your Vcore to go higher than what you set within the bios. But might be needed in your case one day. You could always just try it out once you get this OC half-way solid.

That's weird how more VDRAM made you Bsod o.O There is an explanation for that but lets get this OC stable first :P

I would try out higher LLC just to see how much or if any extra voltage is being applied under load. Normally you want to stay away from 100% LLC but for times when you're going over 5GHz it becomes useful for booting into Windows. Even then it's not always required.

You could just raise the Multi if you wanted. It'd be quicker.

[Wats]

I believe I have the answer.

Update your Bios to 17.7, 17.5 isn't very old at all but it will clear all of the settings out for a fresh start.

Set Vcore to 1.2V

Set Multi to 40

Vdroop offset to 100%

Digital compensation to High

Enable Turbo Boost

Set XMP

Leave everything else Auto

Boot and that should work perfectly. 1.2V is way more than enough for 4000MHz, you should be able to push it to 4.2-4.5 with that voltage.

I wish this forum was bigger.. Hopefully tomorrow will be a better day for OCing.

[rhyseyness]

I believe I have the answer.

Update your Bios to 17.7, 17.5 isn't very old at all but it will clear all of the settings out for a fresh start.

Set Vcore to 1.2V

Set Multi to 40

Vdroop offset to 100%

Digital compensation to High

Enable Turbo Boost

Set XMP

Leave everything else Auto

Boot and that should work perfectly. 1.2V is way more than enough for 4000MHz, you should be able to push it to 4.2-4.5 with that voltage.

I wish this forum was bigger.. Hopefully tomorrow will be a better day for OCing.

Thank you so much. Wish me luck, I'm going for it :P

[rhyseyness]

OH MY GOD, WATS YOU GENIUS! IT'S STABLE AT 1.2 VOLTS!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Words cannot describe how happy I am right now.

Still only at 4GHz, but hopefully I can build up now.

I don't think it's ever been stable at only 1.2V except at stock.

You're amazing. Thank you so much for all your help and perseverance!

[rhyseyness]

Ok, I'm depressed again, although I believe there is improvement...

I'm stable at 1.2V @ 4GHz, but no faster than 4GHz

I tried 4.1 @ 1.2V and BSoD under load

So I put the voltage up to 1.22V, and still BSoD under load, at 4.1GHz

It's like my chip is fine up to 4GHz, but then needs a load more voltage to go up any higher.

This is no doubt improvement, as I said in my last post, I think it's only ever been stable at 1.2V at stock speeds

To hit 4.4GHz, I still needed 1.36Ok, I'm depressed again, although I believe there is improvement...

I'm stable at 1.2V @ 4GHz, but no faster than 4GHz

I tried 4.1 @ 1.2V and BSoD under load

So I put the voltage up to 1.22V, and still BSoD under load, at 4.1GHz

It's like my chip is fine up to 4GHz, but then needs a load more voltage to go up any higher.

This is no doubt improvement, as I said in my last post, I think it's only ever been stable at 1.2V at stock speeds

To hit 4.4GHz, I still needed 1.36V :-/

[Tomm]

With IB, you do need to consider what is important to you: speed or temps. I'm running my 3570k at 4.2GHz but my stable voltage is 1.09v. At full load I'm sitting at 65C (many hours of stability testing). 1.35v is a hell of a lot for any IB chip at 4.4GHz. My advice to you is to start again. Revert your BIOS to stock settings and take a couple of days to work out your overclock. Start with stock speeds, fix your LLC to Extreme or Maximum (varies on each board), turn off PLL overvoltage if its on, set your optimal power phase settings, and manually input your memory settings as they should be (don't use XMP profile), then try finding the absolute minimum voltage your chip requires to run stable @ stock. Once you have your stable lowest voltage, set your first overclock to 4GHz since you know it handles 3.8GHz already. If it runs good with stability testing, push up to 4.2GHz. If you find it runs fine, proceed to 4.4GHz, otherwise, if you get BSOD or any instability issues, notch up the voltage slightly at 4.2GHz until you have it stable. When you do have it stable, try 4.4GHz at that new voltage. Rinse and repeat. I know every chip is different but you should still be able to get a lower voltage at 4.4GHz.

Also, crank up the speed of your fans on the H100i. Mine run at a constant 1600RPM. I'm not particularly bothered about the noise because its minor. Your fans should be spinning very quietly at 1600RPM.

Can I ask, what are your temps @ stock, and what is your case airflow like? And have you tried removing the H100i block and resitting it with new paste?

[firest0rm]

Try putting the fans that came whit your h100i back they are prety good. That should fix the problem.

well then he connected the fans wrong

those fans run at 1500rpm when connected to a molex/3pins power connector

so thats not the fault of the fans really he needs to figure out a way to send more power to the fans

900rpm is to slow

not to mention i run the i5 2500K @ 4.5Ghz + Corsair H50 and i get lower temps

this is definitely either a malfunction or wrongly placed cooler/block

[Barodil]

Maybe you pushed your chip to far and the silicon already degraded so you can't go higher with your overlock. Tom Logan talked about this in one of his videos. Or You just got unlucky...

Btw: What about your fanspeeds? Are they still running at 900rpm? If so, turn them up and stop going over 1,3V! You are going to Kill your CPU and then you achieved nothing but burned money.

[Wats]

Maybe you pushed your chip to far and the silicon already degraded so you can't go higher with your overlock. Tom Logan talked about this in one of his videos. Or You just got unlucky... Btw: What about your fanspeeds? Are they still running at 900rpm? If so' date=' turn them up and stop going over 1,3V! You are going to Kill your CPU and then you achieved nothing but burned money.[/quote']

GL killing an IB with 1.3Vcore :P

No the CPU isn't degraded already, It's just being difficult. Ivy Bridge can be pain at times...

I'm leaning towards he has a very poor chip and might want to delid it just so he can push those extra 0.2Volts, yes 1.5+V. Perfectly safe for IB 24/7.

[Tomm]

Removing the IHS is ballsy. My chip has been good to me so I haven't felt the need but if you have a steady hand and the will to do it, go ahead. New thermal paste under there might actually do the job. The stuff Intel used for the IHS isn't the greatest.

[Wats]

Removing the IHS is ballsy. My chip has been good to me so I haven't felt the need but if you have a steady hand and the will to do it, go ahead. New thermal paste under there might actually do the job. The stuff Intel used for the IHS isn't the greatest.

The glue they used not the paste that causes the heat problems. Remove the glue and have the IHS right on the die with almost no space between then problem solved.

Plus you can use a hammer to delid.. It's kind of new but so far it's safer than the razor method.

[Levent]

Removing the IHS is ballsy. My chip has been good to me so I haven't felt the need but if you have a steady hand and the will to do it, go ahead. New thermal paste under there might actually do the job. The stuff Intel used for the IHS isn't the greatest.

i googled after wats post and i found that was quite interesting

http://www.overclock.net/t/1376206/how-to-delid-your-ivy-bridge-cpu-with-out-a-razor-blade even I might do it right now

[rhyseyness]

Maybe you pushed your chip to far and the silicon already degraded so you can't go higher with your overlock. Tom Logan talked about this in one of his videos. Or You just got unlucky... Btw: What about your fanspeeds? Are they still running at 900rpm? If so' date=' turn them up and stop going over 1,3V! You are going to Kill your CPU and then you achieved nothing but burned money.[/quote']

GL killing an IB with 1.3Vcore :P

No the CPU isn't degraded already, It's just being difficult. Ivy Bridge can be pain at times...

I'm leaning towards he has a very poor chip and might want to delid it just so he can push those extra 0.2Volts, yes 1.5+V. Perfectly safe for IB 24/7.

Sorry, delid?

EDIT: Don't worry I looked it up.... Maybe.

I'm tempted to sell this chip and just buy a new one.

But it is a waste of like £50 :-/

[rhyseyness]

Ok, removing the heat sink seems brave. I could do it at work if I wanted to, (I'm an engineer) but it looks to me like there's a high chance of damaging the chip itself. Does it tend to work? Does it often cause issues?

Just how safe is it to do, in terms of my CPU working afterwards, I guess is what I wanna know?

[rhyseyness]

Not sure if I'm man enough to do it with a hammer :-P

[Wats]

Have you gotten anywhere with the OC?

Even a tiny OC? I think you said 40 w/1.2V worked a day ago or something.

[rhyseyness]

Have you gotten anywhere with the OC?

Even a tiny OC? I think you said 40 w/1.2V worked a day ago or something.

Yea, proper stress tested that oc today (4.0 @ 1.2v) and it's stable. 12 hours of p95.

Haven't tried to go any further yet. I've been on holiday for the last two weeks, but went back to work today, so it's gonna be more difficult to put the time in. I'm gonna see what sort of voltage I need to be stable at 4.1/4.2. See where the real point of temps being unacceptable is...

I've also been thinking about it, and I am really tempted to take off the heat spreader and redo the thermal compound. Like you said, push it to 1.4-1.5V as long as temps stay acceptable. At least get 4.5 out of it... hopefully. I've got some arctic silver sitting around somewhere, and the razor blade method looks safer. Maybe at the weekend

EDIT: hottest core reached 67 degrees at 4.0 and 1.2V