2500k OC Frustration

[dr0be]

I have a 2500k cooled with a H100 in a Gigabyte GA-Z68XP-UD3P. When I first got the H100 (was after the chip had been burned in), I immediately OCd to 4.5GHz. I disabled C1E, speedstep, all that jazz, turned on PLL overload, changed multiplier to 45 and kept getting 0124 BSOD, so I continued to increase vcore and it wouldn't get stable in prime95 blend until 1.375v, which from the infinite amount of threads I read, is pretty high for 4.5GHz.

After a few months, I noticed that at some point it reverted back to the stock multiplier, so I went to change it back to 4.5@1.375v, but it kept BSODing. I got all the ways to 1.395v before it could even clear one pass in prime95 before it would eventually 0124 BSOD per usual - definitely wasn't going to touch 1.4v even if the temps were still okay.

Any idea why this is happening? Crappy chip? Am I doing something wrong? I've been building computers for awhile, so I know the H100 is seated fine (temps are great) - I have just never been OC-savy.

 

[felang]

What temps were you getting at 4.5? I see no reason why you would need to disable Speedstep for a 4.5 GHZ overclock, power consumption will go up while idling.

 

[dr0be]

What temps were you getting at 4.5? I see no reason why you would need to disable Speedstep for a 4.5 GHZ overclock, power consumption will go up while idling.

Barely 50C load in prime95, but I can't even get to 4.5GHz now for whatever reason.

 

[Puppies04]

Why make things so difficult for yourself?

Download and install the latest bios for your mobo and reset your board back to stock. Leave voltage on auto and change the multi to 45. Stress test and if stable start manually lowering the voltage until you bocome unstable then go back up one tick on the voltage. Leave speedstep, C1E, PLL overvoltage etc alone.

 

[dr0be]

Why make things so difficult for yourself?

Download and install the latest bios for your mobo and reset your board back to stock. Leave voltage on auto and change the multi to 45. Stress test and if stable start manually lowering the voltage until you bocome unstable then go back up one tick on the voltage. Leave speedstep, C1E, PLL overvoltage etc alone.

I did this originally and it wouldn't even POST on auto voltage, but I will try again I suppose.

 

[aaksheytalwar]

My CPU 2600k won't be stable at 4.5 either. Just luck.

 

[ShintaiDK]

Yep, no garantees. Just because you read about everyone that gets 4.5Ghz+, doesnt mean _you_ got that.

Abit funny when considering the main arguments just after the IB release...

 

[dr0be]

I installed latest BIOS, reset my BIOS settings, and just changed the multiplier. It wouldn't boot on auto voltage, so I changed it to what the BIOS recommended (1.36v). Opened prime95, hit run, immediately BSOD 0124.

*edit* BSOD 0124 at 1.37v, 1.38v, amd 1.39v on default BIOS settings.

 

[dr0be]

One thing I've noticed that is happening (according to HWMonitor) is that my vcore is dropping below the set value during prime95, which I believe is why I keep BSODing 0124. For example, I'm set on 1.37 right now, prime95 is going fine, and then it freezes and is about to BSOD and the vcore dipped to 1.33.

I turned off all speedstep/C1E/C3/C6 and increased the power limits, but I'm still getting BSOD 0124. Literally have no idea anymore.

 

[Ghiedo27]

How solid are your power supply and motherboard?

Abit funny when considering the main arguments just after the IB release...

To play devil's advocate, I'm sure there's a comparable distribution of people on Ivy Bridge that get really low or unstable clocks for that process also. Bell's curve and all that.

 

[dr0be]

How solid are your power supply and motherboard?

Gigabyte GA-Z68XP-UD3P Rev1.3
Corsair HX750W

More than enough. I bumped the mult down to 42 and I am stable at 1.35v and continuing lower. Not sure why I can't get stable at 45 - even at an absurd 1.4v.

 

[felang]

Gigabyte GA-Z68XP-UD3P Rev1.3
Corsair HX750W

More than enough. I bumped the mult down to 42 and I am stable at 1.35v and continuing lower. Not sure why I can't get stable at 45 - even at an absurd 1.4v.

It seems like your chip is just not a good overclocker, my 2600K does 4.5 at less than 1.3V and it´s far from the best chip out there.

I agree with you that something feels off though... are you messing with base clock at all? Have you tested your RAM thoroughly? Do Gigabyte boards not have LLC?

Also maybe try with PLL overvoltage disabled, shouldn´t be needed for 4.5 ghz anyway.

 

[dr0be]

It seems like your chip is just not a good overclocker, my 2600K does 4.5 at less than 1.3V and it´s far from the best chip out there.

I agree with you that something feels off though... are you messing with base clock at all? Have you tested your RAM thoroughly? Do Gigabyte boards not have LLC?

Also maybe try with PLL overvoltage disabled, shouldn´t be needed for 4.5 ghz anyway.

I'm heading out for a little, but I am stable at x42 @ 1.34v as well. Going to keep going lower after I get back. I've tested all my RAM before when I was trying to figure this out and didn't see any problems. I was trying to OC previously with PLL overv off and I was still getting BSOD 0124.

 

[recycledsole]

i have a 2500k, and an asus p8z68pro gen 3, is there an overclock guide?

 

[guskline]

i have a 2500k, and an asus p8z68pro gen 3, is there an overclock guide?

I used the Asus SuiteII Overclocking tool and it set my parameters to 103fsb and 44 for the clock. You have the same mb I do in rig 1 below. Was easy and painless. That software should be included on the Asus disc.

The Overclocking purists don't like software to do their bidding but, honestly, it was very easy and gave a very nice OC without a lot of tweaking on my part.

I would start there.

 

[BonzaiDuck]

We're missing something here. OP hasn't overvolted the processor enough to do any damage. Is it the PSU? What does the monitoring software (at least) show for the 12V, 3.3V and 5V rails?

How much RAM is he using, and at what voltage? Does it default to 1.65V when the RAM spec is 1.5? Maybe he needs to tweak the VCCIO voltage a tad.

But he should've been able to get to 4.5 on that processor. The use of software for over-clocking would leave him with a voltage surplus, and I'm thinking he should easily get that speed manually at something less than 1.3 -- maybe as little as 1.25 VCORE.

Very strange . . .

 

[poohbear]

OP i think alot of people were misled to believe that 90% of Sandy bridges do 4.5 at almost default vcore. THAT'S NOT TRUE! I had 2 and they needed 1.42v & 1.38v to hit 4.5ghz stable, mind u that's with my mobo the z68 UD3-B3, but i really can't imagine such a huge difference between mobos as it already has a solid 16 phase VRM design. The thing is i was just manipulating vcore, i didnt touch all the other vcore settings, but from my research i didnt seem to need to.

i'm currently doing 4.4ghz @ 1.350. The jump to 4.5ghz @ 1.42v is simply not worth it for me just to get the extra 100mhz.

Regarding your vroop (your vcore dropping under load), there is a setting in your bios that compensates for that and ensures your vcore doesnt drop. Forgot the exact name off the top of my head, but "something" level, just look for the word level in one of the settings and set it to 3,4, or 5 and you'll see the vcore doesnt drop as much under load. That'll definitely help with stability and its the only setting i needed to manipulate aside from vcore to overclock. Initially i had to set the vcore to 1.45 to get it stable, what i didn't realize is that 1.45 was'nt actually 1.45 under load due to vdroop (it'd drop to 1.35), so the level adjustment fixes that. Now i just set it to 1.35 in bios and actually get 1.35 by setting the level adjustment to "level 5". Hope that helps.

 

[coffeejunkee]

Yeah, can't do 4.5GHz without load line calibration. Might have to increase cpu power limits too, so no mobo throttling occurs.

Also, many people seem to think 'auto' means 'stock'. It doesn't.

 

[railven]

OP i think alot of people were misled to believe that 90% of Sandy bridges do 4.5 at almost default vcore. THAT'S NOT TRUE! I had 2 and they needed 1.42v & 1.38v to hit 4.5ghz stable, mind u that's with my mobo the z68 UD3-B3, but i really can't imagine such a huge difference between mobos as it already has a solid 16 phase VRM design. The thing is i was just manipulating vcore, i didnt touch all the other vcore settings, but from my research i didnt seem to need to.

i'm currently doing 4.4ghz @ 1.350. The jump to 4.5ghz @ 1.42v is simply not worth it for me just to get the extra 100mhz.

Beyond my temp issues, I'm finding what Pooh is saying here to be my case. I can get 4.4 @ 1.32 solid with 1hr stress test on Prime95, but I've yet to find working volts for 4.5ghz+, so far I was still BSODing at 1.34v.

Regarding your vroop (your vcore dropping under load), there is a setting in your bios that compensates for that and ensures your vcore doesnt drop. Forgot the exact name off the top of my head, but "something" level, just look for the word level in one of the settings and set it to 3,4, or 5 and you'll see the vcore doesnt drop as much under load. That'll definitely help with stability and its the only setting i needed to manipulate aside from vcore to overclock. Initially i had to set the vcore to 1.45 to get it stable, what i didn't realize is that 1.45 was'nt actually 1.45 under load due to vdroop (it'd drop to 1.35), so the level adjustment fixes that. Now i just set it to 1.35 in bios and actually get 1.35 by setting the level adjustment to "level 5". Hope that helps.

I believe what you are referring to is the Line-In Calibration? The guide I was reading (specifically written for my board which helped) recommends Level 2 or Level 3. I'm running Level 2 currently.

 

[BD231]

1.35v to 1.375v is typical for 4.5ghz to 4.8ghz. Once you go past 4.4ghz your load line setting becomes a much bigger factor in getting the chip stable. Personally I opt to set my voltage slightly lower (1.350v) and up the Load Line which lands me at a much more stable voltage of 1.370v at full load for 4.7ghz. If I lower load line and up the voltage I can't keep the voltage stable at load which makes my chip unstable.

Find someone on youtube with the same board as you and copy their settings, also, disabling C states and speed step is a bad idea. All you need to make sure is off is turbo, everything else will increase the life of your chip.

 

[Kenmitch]

BSOD 0x124 can also be caused by PLL voltage being too high or low for a overclock.

Might be worth playing around with it to see if it helps you or not....For clarification I'm not talking about PLL overvoltage.

 

[Danno21]

I use a M4E-Z motherboard, but for 4.8Ghz, I had to put the vcore to 1.39 and the LLC on high. I left all the C states on and it passed 8 hours of Prime and hasn't crashed yet (over a week so far).

Could be the board and/or the CPU though.

 

[BonzaiDuck]

Wait-a-minute. Is this "fixed-VCore" over-clocking? Of COURSE if he sets VCORE to 1.37V in BIOS, it's going to droop.

I was referring to the drooped voltage.

Sure, his voltage would traverse maybe 1.31 to 1.37V. But I don't have the 2500K -- I have the 2600K, so I wouldn't know the precise numbers . . .

 

[buklau]

Have you try to put LLC to highest as well as reducing the PLL voltage to 1.6v? Normally you won't need to enable PLL voltage until you get pass 4.5ghz.

 

[FalseChristian]

I just changed the 'By all cores' to 45 and left the core voltage on 'AUTO" and it's stable as hell at 4.5GHZ using 1.275 volts.