Overclock.net › Forums › Intel › Intel CPUs › [Solved] I5-2500k Overclocking help
New Posts  All Forums:Forum Nav:

[Solved] I5-2500k Overclocking help

post #1 of 35
Thread Starter 
I really detest the fact that I have to create a separate thread for what seems to be an endless supply of the same topic over and over again, but alas I see no better option. Rather than derail, or hijack other threads, I found it best to create my own thread, even if it is a single thread in a sea of the same thread type.

I have read the overclocking guide for Sandybridge, and I have read the one specific to Asrock boards, however I have not read the entirety of either. I find I learn best through hands on experience, and sometimes that requires a little help along the way. I have found a baseline so far, in which I have achieved at the minimum 1 hour of blend stability at 90% ram utilization. The settings I am currently using are 45x multi, 1.32v vcore, 1.72 PLL and 1.1v Vccio. My ram is set to the defaults, and I have zero issue with stock settings.

When I started my journey, it was a couple days ago when I finally received the bracket kit required to mount my Xigmatek Dark Knight to my CPU. I started off at 45x multi, 1.36v vcore and default manual voltages for PLL and Vccio, which were 1.08 and 1.5 I believe. I was able to boot into windows once, and shortly after I did I got a BSOD, error code 0x124. I read that this code could be caused by insufficient Vcore, or Vccio. So I increased my vcore to 1.4v to no avail. I found that my hard disk had been corrupted and I could no longer boot into windows, even with stock settings.


After running chkdsk, I was able to repair the data corruption, and resume my quest. I noticed that in windows on my first attempt that prior to the BSOD, I was showing ~1.26v in CPUz. I figured that the excessive Vdroop was causing my instability, so I decreased my Vcore to 1.32, and set LLC to level 1 ( the highest setting). I also set Vccio to 1.1v (which seems to be standard for higher clocks) and my PLL to 1.72v as a precaution. With these settings I was able to boot into windows and run prime for over an hour with no issues before I stopped the program.

I tried lowering my Vcore to 1.3v just to see if I could remain stable. I was able to boot into windows, and was able to run 15 minutes of prime before I got the 0x124 code again. This leads me to believe that my Vcore is the culprit, but the fact is I know very little about the Sandy architecture in regards to overclocking, which is why I have created this thread.

I seek assistance in determining the best settings,and a stable overclock. I have experience with overclocking, and have overclocked three C2D since with great success, Sandy is a bit different. The methodology seems the same, just the components and settings seem to be an entirely different animal.

I would like to achieve at least a 4.5ghz overclock with a minimum of 12 hours stable in prime. The max temp I am willing to accept at load in Prime is 80c. I have some headroom, but being that my cooler isn't the most efficient cooler on the market, I am somewhat limited.

Thanks in advance for your guys help!

TL;DR

Current settings are
1.32v vcore
1.1v Vccio
1.72v PLL
~1 hour stable in Prime with 90% ram

I would like to achieve 4.5ghz minimum, under 80c, 12 hours prime stable at least.
Edited by Mattb2e - 8/15/12 at 1:20pm
post #2 of 35
mmm E3G3, I love you.

I will grab a snapshot of my settings to share with you on my 5ghz OC.

Edit: First, see if using the other 2 ram slots help, I have the best overall stability in the channel pair furthest from the socket.
post #3 of 35
Thread Starter 
Quote:
Originally Posted by Xinoxide View Post

mmm E3G3, I love you.
I will grab a snapshot of my settings to share with you on my 5ghz OC.
Edit: First, see if using the other 2 ram slots help, I have the best overall stability in the channel pair furthest from the socket.

I love the E3G3 as well, its damn sexy biggrin.gif

I am using only two slots, and they are the furthest from the socket, just as you have done.

Being that I BSOD'd from reducing Vcore, I would think its something else other than ram. Then again, it used to be blend was used to stress the MCH, now that the memory controller is part of the die, it changes things quite a bit. When I get a bit further into a baseline I will see if swapping ram slots changes anything.
Edited by Mattb2e - 8/10/12 at 8:31pm
post #4 of 35


Not sure if it will help.

I have heard that Level 2 is more useful for high clocks like this, but I just came down from trying for 5400mhz. So that rule got lost in translation. Forgot I just reloaded the profile I had.

When I first got the board, I was SOO surprised it could do this.
post #5 of 35
http://hardforum.com/showthread.php?t=1578110

http://www.youtube.com/watch?v=9e04-5hjk-I

On some of those boards..the only LLC that works is 1 ..can't remember which..I made the mistake of
buying several within a short period of time.They're all dead now rolleyes.gif
They boot fast! smile.gif
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
post #6 of 35
I will then do some further testing and researching into this.

I really thought that rule was only for suicide runs.
post #7 of 35
Thread Starter 
Quote:
Originally Posted by Schmuckley View Post

http://hardforum.com/showthread.php?t=1578110
http://www.youtube.com/watch?v=9e04-5hjk-I
On some of those boards..the only LLC that works is 1 ..can't remember which..I made the mistake of
buying several within a short period of time.They're all dead now rolleyes.gif
They boot fast! smile.gif

You killed all your boards? How did you do that?

For whatever reason, LLC is really my only option, aside from setting my Vcore to 1.5v in the UEFI. If my board didnt have so much Vdroop and Vdrop, I wouldn't need to use it. It seems I need somewhere around 1.32v in order to be somewhat stable at 4.5ghz. I was able to boot at 5ghz at 1.4v, but BSOD'd 0x124 as soon as windows got to the desktop. I bet that 4.7ghz or 4.8 would work at 1.4v. Then again, im not to sure I want to run my voltage that high, until it gets a bit older and the prospects of an upgrade are on the horizon biggrin.gif.

Thank you for the links, the HardForum link is most useful, and provides me with some pretty detailed information. I was unaware that my PLL voltage was too high, and that lowering it could reduce temps a bit. So im now running a lower PLL voltage.

I am still a bit confused with the whole offset voltage settings. I need to use a negative offset to achieve the proper Vcore at load. Should LLC be set to minimum for the offset Vcore to work properly?

Quote:
Originally Posted by Xinoxide View Post

Warning: Spoiler! (Click to show)
Not sure if it will help.
I have heard that Level 2 is more useful for high clocks like this, but I just came down from trying for 5400mhz. So that rule got lost in translation. Forgot I just reloaded the profile I had.
When I first got the board, I was SOO surprised it could do this.

You must have a golden chip then, according to the statistics in the link that Schmuckley provided, only 10% of SB chips can achieve a 5ghz clock, im sure 5.4ghz is an even smaller percentage. Then again, booting at said frequency is miles from being stable biggrin.gif.

As far as LLC goes. Ive been dead set against it for years, because of its ability to damage the CPU with voltage spikes. While generally it wont damage your CPU if your not running suicide voltages, I never really liked the idea of voltage spikes, even if its within the VID range of the CPU. Unfortunately, on my C2Q rig, I was spoiled with a board with somewhat decent voltage control, so my Vdroop wasn't nearly as severe as it is on this board.

From every guide ive read thus far, offset voltage is the way to go, I just gotta figure out how to configure the proper setting. If that settings includes disabling LLC, I would be more at ease as well.
post #8 of 35
Quote:
Originally Posted by Mattb2e View Post

You must have a golden chip then, according to the statistics in the link that Schmuckley provided, only 10% of SB chips can achieve a 5ghz clock, im sure 5.4ghz is an even smaller percentage.
10%... that is hogwash. I have only seen a handful of chips that couldn't do AT LEAST 50, a large percentage of those stable up to 1.52v ( highest I would go with air ). ITs gotta be more than 10%.

This board does have a quite a bit of droop. However I found the LLC to be somewhat stable.
Edited by Xinoxide - 8/11/12 at 4:34pm
post #9 of 35
Thread Starter 
Quote:
Originally Posted by Xinoxide View Post

10%... that is hogwash. I have only seen a handful of chips that couldn't do AT LEAST 50, a large percentage of those stable up to 1.52v ( highest I would go with air ). ITs gotta be more than 10%.
This board does have a quite a bit of droop. However I found the LLC to be somewhat stable.

IDK, thats what the guy from Asus said lol.

LLC seems to be stable, and it works, but I am not too crazy about the thought of voltage spikes from using it. Any idea how I should setup offset vcore?
post #10 of 35
You just increase the offset anytime you get a 101/124 bsod from stepping up the multi. :]
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › [Solved] I5-2500k Overclocking help