Sign in with
Sign up | Sign in
Your question

Frustrating 3770k OC Results - Suggestions?

Last response: in Overclocking
Share
December 26, 2012 6:43:20 AM

Brand new 3770k build:

Asus P8Z77-V Pro
H100i - 2 fans, both in pull
GPU cooled with an old H50
Thermaltake 850 Watt PSU

I've spent the past several days trying to get a decent OC on this rig with, frankly, really frustrating results. I've read a bunch of literature describing people on the extreme end of the spectrum getting up to 5GHz on slightly higher than stock voltages and air cooling. The less-extreme-but-still-good chips seem to get up to 4.8GHz on somewhere around 1.35v, maybe slightly less. Regardless, it seems that nearly everyone that wants it can access a 4.8GHz OC with decent cooling pushing 1.38 core voltage.

I'm currently struggling HARD to get a stable 4.6GHz OC with a .055v offset!! That means that under the right conditions, LLC and the board AI (I guess?) are feeding the chip close to 1.4v - and the damn thing STILL can't pass a rigorous round of ITB testing. With that kind of voltage on this chip, the stress testing process has the hottest core pinned up against TJmax - or a handful of degrees shy of it most of the time - even with the H100i's dual 2500RPM fans screaming so loud I'm convinced the friggin thing is about to start hovering.

My proc just failed another round of heavy ITB testing, so I've bumped the offset up to .055. At that level I can no longer use ITB. The first 5 minutes of testing had the hottest core peaking at 97c :sarcastic: 

So that's it, then? 4.5 is really the best this thing can get? Glad I invested in this expensive aftermarket cooler...I've read of some people getting that on the stock intel cooler. :fou: 

I've GOTTA be doing something wrong, right? At this point I'd LOVE to be doing something wrong...

a c 194 à CPUs
a c 125 K Overclocking
December 26, 2012 11:22:37 AM

OK start over with FIXED voltage. Set your V to stock and see how far you can go on the multiplier. Offset is a horrible way to OC if you don't know what the chip can do in the first place. I always recommend fixed at first.
December 26, 2012 3:58:58 PM

Novuake said:
OK start over with FIXED voltage. Set your V to stock and see how far you can go on the multiplier. Offset is a horrible way to OC if you don't know what the chip can do in the first place. I always recommend fixed at first.


Thanks for the reply.

Yeah, that's how I started. I was seemingly stable with at 4.6 with 1.328 vcore. That was with LLC doing its thing. I think I had the voltage manually set to 1.31v
Related resources
a c 194 à CPUs
a c 125 K Overclocking
December 26, 2012 4:10:12 PM

Ok what LLC have you set? Set it to lvl 2 or even lvl 1...
December 26, 2012 5:03:44 PM

I have LLC set 75% of max. "Ultra High", in Asus terminology.
December 26, 2012 6:01:47 PM

Update:

CPU failed after an 1.5 hours of Prime95 at .055 offset. BSOD, hard reboot. Guess it's time to go up to .060, but that's really pushing the voltage up there...
a c 194 à CPUs
a c 125 K Overclocking
December 26, 2012 6:08:22 PM

Hang on... Don not push it so much. Seriously, set fixed to high and then low to see which one is the most rigid. In fixed mode..
December 26, 2012 6:09:35 PM

Alright, I'm willing to try anything at this point...haha

But what exactly do you mean by setting fixed to low?
a c 194 à CPUs
a c 125 K Overclocking
December 26, 2012 6:36:25 PM

OK LLC(load line calibration) regulates your Voltage to a close degree of the fixed voltage. So, what I am suggesting is that you set that allowed fluctuation to as little as possible. On my board LVL 1 is the most rigid, but I am not sure if High or Low translates to that on your Asus board.

So figure out with CPU-Z which is the most rigid, low or high(or percentage) adn then set it to the most rigid you can get. Start at 1.2V and see what the max multiplier you can reach.

You have also turned of all power saving and turbo features right?
a b à CPUs
a b K Overclocking
December 26, 2012 6:52:32 PM

use fixed mode. you can get a lot less vcore and still be stable. at least i can. like .05 less and it will be cooler. i think you have miss mounted that fisher price water cooler though. those temps seem too high.
a c 184 à CPUs
a c 149 K Overclocking
December 26, 2012 7:22:56 PM

Fixed. Offset is really a trial and error method....
December 27, 2012 4:17:39 PM

cbrunnem said:
use fixed mode. you can get a lot less vcore and still be stable. at least i can. like .05 less and it will be cooler. i think you have miss mounted that fisher price water cooler though. those temps seem too high.


I've mounted and remounted it several times. No change. In fact, my temps are just getting worse. I had to change out my motherboard because my old one stopped recognizing when things were being plugged into it. The highest temps on original install were 82 93 89 86. I changed out the thermal paste to Nano Diamond Formula 7, now I'm looking at 86 98 90 86
a c 194 à CPUs
a c 125 K Overclocking
December 27, 2012 4:51:06 PM

jazzmac251 said:
I've mounted and remounted it several times. No change. In fact, my temps are just getting worse. I had to change out my motherboard because my old one stopped recognizing when things were being plugged into it. The highest temps on original install were 82 93 89 86. I changed out the thermal paste to Nano Diamond Formula 7, now I'm looking at 86 98 90 86


I am starting to think you have a defective pump on that H100...

What are your stock temps?
December 27, 2012 5:55:22 PM

Novuake said:
I am starting to think you have a defective pump on that H100...

What are your stock temps?


You think? I don't know about that... I have to send it back anyway because the fan controller stopped working. Currently running the fans off the mobo. I'm pretty sure the pump still works though.

Idle temps are usually under 30c
December 27, 2012 6:39:52 PM

I have noticed something odd during my ITB runs. I'm burning at the maximum setting. As each run nears completion and during the first minute of so of each new run, my processor temps are much more reasonable, usually around 70c. Outside of those periods, my temps hover around a sustained 94c. What's strange is that I'm watching AIDA64's CPUZ, and nothing is changing between those time periods. Voltage is the same, CPU load is still pinned at 100%, clock speed is the same. How can there be a consistent 25c delta between two periods of max CPU usage? What gives? Is that just the magic of Linpack?

Update: Idle temps are around 35c with a fixed 1.32 CPU voltage.
!