i5 3570K Overclocking Limits Dropping Over Time?

Toggle sidebar Toggle sidebar
D

dpshelton

Distinguished
Jan 5, 2012
11
0
18,510
I built my system about a year ago:

i5 3570K
ASRock Z77 Extreme4
G.SKILL Ripjaws X Series 8GB
XFX Double D FX-785A-CDFC Radeon HD 7850 2GB

When I first built the system, I could overclock to 4.4 GHz, no problem, with +20mV offset. After 2 months, games began crashing unexpectedly (the OS never crashes, just the game...). So, I dropped the multiplier to 43 (4.3 GHz). This worked for another 3 months, then the crashes began again. Over time, my overclock limit has seemed to drop every few months. I'm now limited to 4.0 GHz. Ramping up the offset any more than +20mV has never allowed me to increase my multiplier any more for this processor (bad luck I guess...).

I never had to reduce my overclock on my Core 2 Duo machine (still running smooth and fast after 7 years). So, I'm a little confused about this long-term issue I'm seeing. My temperatures are below 65 C under my heaviest load, and I don't overclock my video card. Has anybody else had this issue and fixed it? Is there a setting I'm missing (besides core offset and multiplier)? Could it be the RAM and not the CPU? I can provide all my mobo settings if needed.

Thanks.
 
Solution
B
well, whatever you do 1.4v is reasonable upper limit, 1.5v is the wall - DO NOT exceed. but of course lower the better. especially with that cooler - not a bad cooler, but really only designed for mid end OC. Keep an eye on temps during load stress testing. IF you begin exceeding 80c on load stress test, or 70-2c on normal daily use, either back off volts and OC, or upgrade cooler
Sort by date Sort by votes
B

Buzz247

Honorable
Mar 18, 2013
962
1
11,360
k1114 is on the right track - this behaivior indicate failing PSU supplying dirty power, or CPU damage has occurred from overvolting at somepoint, or CPU damaged occurred from overheating at somepoint. Essentially, sounds like you may have CPU transistors failing.

There are a couple of other avenues possible:

have you actually monitored your voltage values using Hardwaremonitor.org or something like it? You may have a failing VRM on your motherboard. watch for massive fluctuations of power on your CPU feed rails. even 1-2% can cause this scenario.

You listed your CPU package temp, but have your used anything like Real Temps to get your actually core temps? if one core is significantly higher or lower than others, there is a CPU problem.
 
Upvote 0 Downvote
D

dpshelton

Distinguished
Jan 5, 2012
11
0
18,510
Thanks for the replies. Here is some additional info...

Mobo Settings
CPU Ratio: 40 (currently... I had it at 44 when I built the system)
Host clock override: 100 (default)
Additional Turbo Voltage: +0.027V
Offset Voltage: -0.010V (for low-power idling)

Here are some stats after running prime95 for about 30 mins. From my somewhat untrained eye, things don't look abnormal (again, untrained eye...):

0mvWHU0.png


The funny thing is that it seems to occur only during graphics-intensive applications, so I would expect the issue to be the GPU. However, the crashes go away after reducing the CPU's frequency, not the GPU's... Also, it's just the game that crashes. Nothing else. No blue screen, etc...

Thanks for the help guys!
 
Upvote 0 Downvote
B

Buzz247

Honorable
Mar 18, 2013
962
1
11,360
Well nothing looks wrong per se - but I am rather confused now. Has your Vcore always been at 1.26? normal for that CPU at stock value is 1.18-1.25v. Above 4.0 (sometimes even at 4.0) you can expect to have to increase vcore to compensate for increased cycles. Normal of 4.4 would be around 1.3-.33. Now of course, every chip is different, so some variations in vcore are expected, but those are general expectations.

also have you tried kicking your GPU back to stock values and increasing your CPU OC to rule out a gpu issue caused by drivers, power, vrm or temps?
 
Upvote 0 Downvote
D

dpshelton

Distinguished
Jan 5, 2012
11
0
18,510
Buzz.
Yep, Vcore has always been around 1.26. I just upped my ratio to 4.3, and it's holding steady at 1.26V.

Also, I keep the GPU at stock values ever since I first noticed the issues, but I still see them.

Anyways, if nothing sticks out to you all about my stats, maybe I just need to stick to a lower overclock and hopefully the limit will even-out instead of steadily decreasing. Thanks for the help guys.
 
Upvote 0 Downvote
B

Buzz247

Honorable
Mar 18, 2013
962
1
11,360
ok - here's a remote possibility - you had a fluke. that kind OC on that low of voltage is called a "golden chip". Rare to be sure. a number of things can cause this to happen, from a perfect set of transistors, and over performing core, a perfect pairing of die to IHS, etc etc etc. It is rare they lasted. What you are now experiencing is actually normal and expected values for that chip. If you want, you can try to up the vcore in steps, following normal OC procedures of up, boot, test, rinse n repeat. That very well may get you back up where you were, again at normal expected performance values. I say all of this assuming you have an aftermarket cooler on there of course, not the stock intel cooler that came with chipset.

Still, try the gpu flip that i suggested above - you may have a problem with the gpu and not even realize it because of this odd behavior

edit: nevermind the gpu thing - drop cpu back to stock - try normal use at stock - see what happens. I really am thinking you had a fluke that never was truly stable at that voltage, and it is simply rearing it's head now for whatever reason. if everything is fine at stock, beging OC procedure, increasing vcore as needed - make sure to test on Prime95 at each step to insure stability before proceeding to next multiplier. prime95 is where volt tweaks can often show right away
 
Upvote 0 Downvote
D

dpshelton

Distinguished
Jan 5, 2012
11
0
18,510


Sounds like a good plan. I will try to re-calibrate my OC from the beginning. I have a True Spirit 120 CPU cooler, so hopefully I can keep the temps reasonable. Is there an upper-limit to the voltage I should try to stay below? You mentioned 1.3-1.33. I'll try to stay around or below that. Thanks again!
 
Upvote 0 Downvote
B

Buzz247

Honorable
Mar 18, 2013
962
1
11,360
well, whatever you do 1.4v is reasonable upper limit, 1.5v is the wall - DO NOT exceed. but of course lower the better. especially with that cooler - not a bad cooler, but really only designed for mid end OC. Keep an eye on temps during load stress testing. IF you begin exceeding 80c on load stress test, or 70-2c on normal daily use, either back off volts and OC, or upgrade cooler
 
Upvote 0 Downvote
Solution
thequn

thequn

Distinguished
Nov 4, 2012
393
0
18,810


Dude you just answers a question i have been trying to find and answer to for 3 years that i just found by mistake in January but didnt know it was the issue i love you.
 
Upvote 0 Downvote
D

dpshelton

Distinguished
Jan 5, 2012
11
0
18,510


Thought I would update you. I re-calibrated my overclock, and you were absolutely right about the CPU not being stable at the Vcore voltage. Even though it passed prime95 at that voltage, it must not have been 100% stable. I now have my voltage at around 1.28 (used to be 1.26) at 4.4 GHz, and programs aren't crashing anymore. Now I'm tempted to push even higher since I still have good temps :)

Thanks for your help!
 
Upvote 0 Downvote
D

dpshelton

Distinguished
Jan 5, 2012
11
0
18,510


When I first overclocked, I ran prime overnight (8+ hours). This time, I ran it for only 30 mins. I'm realizing that my true test is within Starcraft 2 (strangely enough, the "buggy" scenario shows up in SC2 and not in Prime95...).
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom