Complete Overclocking Guide: Sandy Bridge & Ivy Bridge | *ASRock Edition* - Page 127

It definitely varies per chip. The concept of overclocking is that we are using up margin that Intel has in their system to account for variability in their manufacturing process and in order to stay within a thermal and power envelope. As with any real life manufactured item, you don't really know how much margin you have until you fully test it. Intel does some testing and does that to bin chips into the various categories, but it is still only rudimentary testing.

If you followed the guide, then you walked your way down until you were stable enough to boot and then walked your way back up to the 4.8 GHz. You may have noticed a few things as you were walking your multiplier back up. First you probably noticed that you were having to increase your turbo boost core voltage. The error messages from blue screens of death of windows give you a hint as to what each value had to be changed if you followed the notes of what they mean. You likely have also had to increase your VCCIO/VTT/QPI voltage, possibly your RAM voltage and possibly had to increase your RAM timings to get stable. For me 4.8 was the real start of the test, so you may not have tried some of these.

I cheat a bit and use IntelBurnTest on "Very High" since it does very well at causing the early crashes to happen very fast. It doesn't do all of it, but it helped me get closer quicker than just Prime95 alone.

Is that the hottest your cores are getting? Personally I have Coretemp running at all times since it records the maximum temperature in addition. I've kept mine below 80 as best as I could. Your Vcore is higher than mine is at 4.8, but it doesn't look completely out of bed. I'm guessing you are air cooling.

The way that I've got where I am is careful methodical adjustments and paying attention to every blue screen error. I was using a spreadsheet to detail what I changed each time, but a friend (one of the LN2 extreme folk) just uses a notebook to keep track as he goes along. I have been noting "trends" as I go along so I can make a guess as to how much of a change I need to make rather than just randomly changing a value. You might be clocking too high for how much cooling you have available. If you haven't done it, you should try to get stable at 4.7 or perhaps 4.6 to see where that puts you for temperatures as the voltages will be slightly higher for 4.8 and the temperatures will be higher as well. When I got the screen freezing without blue screening, it was a mixture of RAM timing, VTT/QPI/VCCIO, and RAM voltage. I upped the RAM voltage a bit along with the VTT and that was enough to get a blue screen which indicated that the timing/ram voltage/vtt were too low.

What was the last set of BSOD messages you got?

Can you show us your actual settings page as well as the monitor page?

Im having a new problem with my system. I would be very thankful if someone could help.

When overclocked Im getting total system shut down under heavy load, at first I thought it was my old OCZ Powerstream 520 but after recently replacing It with a new one the problem persist.

Heres my stress test methods:
CPU is over clocked to 4.4ghz. Prime test (Using custom setting from OP) for near 24 hours no errors or crashes. Temps are 30c idle 86c load (only 1 core hit 86c for 2 seconds the rest were low 80c)
Memtest the memory over 2 nights with no errors.
MSI Kombustor GPU burn in test over night with out crash. Temps hit 72 max at load

After all those test I figured I was stable until I went to play Guild Wars 2 beta stress test. A few hours into playing, my PC just shut down like someone hit the switch then after about 10-20 seconds it turns back on automatically . I booted back up with no problems started playing again then 20 min later it happened again. So I went into the bios and set it to default and and continued playing with out any problems.

After the stress test was over I reran all the test and everything passed.
That got me thinking that it must be when the CPU and GPU are stressed at the same time, So I ran prime and GPU burn in at the same time and after about an hour it shut down.

Anyone know why this would be happening?

Heres my Bio settings
CPU Config

---

CPU Ratio: all cores
all cores: 44
host clock override: 100.0
spread spectrum: disabled
intel speed step: disabled
intel turbo boost : enabled
additional turbo voltage: +0.098v
internal pll overvoltage: disabled
long duration power limit: 500
long duration maintained: auto
short duration power limit: 500
primary plain current limit: 500
secondary plain current limit: 500
gt OC support: disabled

Voltage Config

---

power savings mode: disabled
cpu voltage offset : offset mode
offset voltage: +0.015v
cpu load line calibration: 100%
igpu voltage offset: auto
igpu load line calibration: auto
dram voltage: 1.500v
vtt voltage: auto
pch voltage: auto
cpu pll voltage: auto
vccsa voltage : auto

If you need anymore info please let my know.

it my be CPU temperature! Pay attention -> not a core temp. Probably that air flow in your case is not good enough

You may need a higher offset voltage. Also check to see if your RAM settings are set to their rated settings and not set to Auto.

Thank you very much for this guide! Extremely helpful, even though I have done overclocking before.

My 3570k got a surprising overclock on my Z77 Pro 4 - 4.6Ghz on stock voltage, 23C idle and 63C load (obviously the lower number is so low due to speed-step). So long as my chip is cool and has a low voltage setting it should last a while, right?

Here is to hoping!

Grats on your good chip! It should last you for many years as overclocking on stock voltage does not really lower the lifespan of a component compared to overvolting. Even with overvolting, there are still users using chips from 5 years ago that are still running strong.

My gf's Q6600 has been overclocked & overvolted on a crappy aircooler and has not presented any problems in the 4 years it's been running and she normally has a very dusty computer.

I just recently got a 3570k when I built my new computer and currently its clocked at 4.4 ghx @ 1.23v with +.005 offset and +.004 turbo. These are my desired clocks for the time being and was looking into undervolting to reduce heat and increase longevity. I know your guide reccomends against this but other people around here have had theirs clocked at insane voltages at my OC like 1.15v. How should I go about doing this and is this recommended?

Hey!

First of all, thanks for the helpful guide! It's my first time overclocking and the guide has helped me a lot. However, I've sadly run into some issues (I think?).

My hardware:

ASRock Z68 Pro3 motherboard
Intel i5 2500K @ 3.3 GHz CPU

So, here's the situation: At the moment I'm trying to get a stable overclock at 4.4 GHz. I worked my way up there from 3.3 GHz, increasing the multiplier by 2 after each succesful boot and 5 minute Prime95 test with 0 errors and 0 warnings. However, I've now noticed that after a couple of minutes of the Torture Test, the multiplier displayed by CPU-Z and RealTemp suddenly drops from 44 to 16 for a few seconds, and then shoots back up to 44. For this to happen the first time during a test it takes usually about 2 minutes or so, and after the first drop, others come every now and again. My temps during the testing are most of the time floating around 70, max recorded temp during testing being 76, way below the optimal 85 mentioned in the guide. So my question is: is this actually an issue? And if yes, ... Halp? Any ideas on fixing it would be greatly appreciated!

My BIOS settings:

CPU control
Max ratio: 44
Internal PLL Overvolt: Disabled
Intel SpeedStep: Disabled
Core Current Limit: 300 (max)
BCLK: 100.0 MHz
Spread Spectrum: Disabled

Voltage
CPU Core Voltage: Offset mode
Offset Voltage: +0.005 V
CPU Load-Line Calibration: Level 5

Pretty much all other BIOS settings are on Auto.

Is SpeedStep enabled or disabled going to be more likely stable?

Also... what about WHEA's? Is increasing VCore the only thing to do about WHEA's (I get code 19 parity errors, always on the same core--curiously, the coldest core) or is there something else that can help resolve them?