I5-4670K overclocking advice

[Dr. G]

Hi everyone,

 

I'm a long-time Linus Tech Tips follower, first time poster  

 

I finally upgraded my old socket AM3 system to a LGA 1150/Z97 board, and I'm learning about overclocking Intel systems. 

 

I'm running a 4670K processor, and I've already had pretty good luck with overclocking. So far, I've been able to hit 4.4ghz with a vcore of 1.25, and my temperatures during x264 stress tests max out at 72C, running a 212 plus air cooler. At 4.5ghz (1.25vcore) the system will BSOD after just a few minutes of stress testing. 

 

I understand that Haswell CPU's start to heat up rather quickly past 1.25v.....but i'm curious about increasing my vcore slightly, and seeing if I can eek out another 100mhz or two. Given that i'm only hitting low 70's in temperatures, I figure I have a bit of thermal headroom....as I understand 80C is OK for a 24/7 system. 

 

Any advice or suggestions would be helpful. 

 

Thanks. 

 

 

[natey_g]

1.3v is the max I would go, depending on your mobo.

[GER_T4IGA]

80°C would probably be OK for a chip but permanently a little bit to much stress for the hardware if you ask me. What are you doing to it in every day use? You might never hit even 70 with your 1.25VCore in every day use?

[aisle9]

At 72C, you have a lot of headroom left. I have a hard cap for my 4790K at 85C during synthetic testing (OCCT, AIDA64, RealBench). When you overclock, two numbers are important: the highest voltage you're willing to go to, and the highest temperature you're willing to hit in synthetic stress testing. Clearly, you've got room in both categories (1.3v is considered the practical max for Haswell outside of benchmark runs).

 

Click up your voltage in tiny increments until one of three things happen: you're stable at the next 100MHz mark, your temperature hits the maximum you're comfortable with, or you've put as much voltage through the CPU as you're willing to and are still BSOD-ing. At some point, you'll either temperature or voltage out, and on a 212 EVO it'll probably be temperature that puts you out of the game. I have a "guide" of sorts in my profile that details my methods for overclocking.

[Dr. G]

Thank you for all of the replies!

 

I did a little more testing tonight, and it appears that 4.5ghz might be my best OC on this chip/cooler setup. 

 

4.5ghz was seemingly stable at 1.29v....but I need to do longer stress testing. 

 

4.6ghz was only stable at 1.39v....but my temperatures quickly hit 86C, and I had to abort the test. 

 

I'll be doing more stress testing at 4.5ghz, and I will see if that is stable. If not, I think I'm going to be content with 4.4ghz, and start lowering vcore until I am no longer stable. 

[GER_T4IGA]

On 6.9.2016 at 6:35 AM, Dr. G said:

Thank you for all of the replies!

 

I did a little more testing tonight, and it appears that 4.5ghz might be my best OC on this chip/cooler setup. 

 

4.5ghz was seemingly stable at 1.29v....but I need to do longer stress testing. 

 

4.6ghz was only stable at 1.39v....but my temperatures quickly hit 86C, and I had to abort the test. 

 

I'll be doing more stress testing at 4.5ghz, and I will see if that is stable. If not, I think I'm going to be content with 4.4ghz, and start lowering vcore until I am no longer stable. 

Disclaimer: This might not be the best thing to do. It is my own experience and how my OC turned out.

 

I never did any extensive stressing other than real world use as in using my system . I have a handful of crashes after trying to reduce my voltage but I never really tested with AIDA Prime or something. I just gamed away and everything was fine 4.6@1.3V BUT I also don't have any critical stuff on my system and the worst thing I might lose from my system is a game's savedata but no video projects or something so I am not the best place to ask for a stress tests use.

[Dr. G]

Well folks, this CPU overclocking journey has come to an end. After about a week of overnight stability testing at various speeds/volts, I believe I have found my optimum OC. 

 

4.4ghz @1.24v is about all that this chip will do.

 

I thought I might be able to squeeze out another 1-200mhz with voltage increases, but no dice. 4.5ghz @1.37v would not yield 8hr x264 stability....rather it would crash after about 4-5hrs. At this voltage I was also hitting 85C max core temps. 

 

I also experimented with raising my VCCIN to about 1.95v....that didn't seem to change anything....nor did manually setting CPU LLC to "high". 

 

Looks like I have about an "average" 4670K.

 

Overall I am quite happy - my single and multi-core benchmark scores in Cinebench 10 are very close to double the performance of my old Phenom II 980.

 

 

[Dr. G]

Well, I've been fiddling around with this chip more....

 

Just for fun I tried increasing my VCCIN to 2.0....and I've found stability at 4.5ghz @ 1.35vcore. 

 

I'm running an Asus Z97-E mobo....I suspect that the higher VCCIN is preventing vcore drops during stability testing. 

[GER_T4IGA]

7 hours ago, Dr. G said:

Well, I've been fiddling around with this chip more....

 

Just for fun I tried increasing my VCCIN to 2.0....and I've found stability at 4.5ghz @ 1.35vcore. 

 

I'm running an Asus Z97-E mobo....I suspect that the higher VCCIN is preventing vcore drops during stability testing. 

To be fair I have no idea what all the voltages do but I have to admit 2V is pretty high for anything. You're sure you are safe there?

 

[Dr. G]

4 hours ago, GER_T4IGA said:

To be fair I have no idea what all the voltages do but I have to admit 2V is pretty high for anything. You're sure you are safe there?

 

From what I've read in multiple Haswell overclocking guides, 2.1v is the max safe limit for VCCIN.