Kernel_task is using over 600% of my CPU and slowing things down to a crawl!!!

As annoying as it sounds, your use case, is actually the way Kernel task is meant to work... For everyone else on this form who had the Macbook8,2 or older this would happen regardless of rendering or processing at all.

It really sounds like the Newer thiner Retina MacBooks, may have a really hard time dispersing heat.. If this is the case, I would NOT modify the plist because there are other parts of the computer that can break or even explode if heat is not dispersed properly. Its thinner too, making that all the more difficult..

Clean out out your fans.. • Difficulty: Difficult

http://www.ifixit.com/Guide/Replacing+MacBook+Pro+15-Inch+Retina+Display+Mid+201 2+Left+Fan/9719/1

Let me suggest, compressed air is really ineffective on the squirrel fans, in the Mac line up.

I messier about 1cm Of dust buildup per year in the Macbook8,2.. Your going to need a person who good at opening electronics, to remove the fans clean them and reinstall them.
This also makes the computer completely silent.. Like it was out of the box.

You could be right. On the other hand:

1) It worked. kernel_task utilization went from 200+ to <10% and has stayed there for two days under all operating conditions. And the Macbook is completely silent.

2) Fan noise is determined by fan speed and air flow, not by fan cleanliness. If the operating temperature is kept lower by more effective air flow, then the fan does not have to work as hard. The fan speed will be lower and the intake air velocity lower, both leading to much less noise. The noise is a key indicator; if the noise is high then the temperature is high and the airflow is less than what the system desires.

In case this helps anyone...

My Late 2012 iMac 27", 3.4GHz i7, GTX 680MX, developed this issue. No entry in plist for this machine either, and the 'remove all plist entries' fix had no effect on the problem for me.

Luckily I had actually purchased the machine in May '13, so it was still under warranty. Worked with Apple Support online, and when Hardware Test failed three times in a row (got to 16 seconds tested, hung for a few min than shut down each time) they agreed it was a hardware problem.

5 days, a new fan and new logic board later (all covered, thank goodness), machine is back to normal.

Moral of the story is that, at least for me, the root cause *is* a hardware issue.

I ran the command :kextstat -kl | awk '!/com\.apple/{printf "%s %s\n", $6, $7}'

Then removed the one program that prompted with that command, restarted and now my kernal_task real memory jumped down from 3 GB to 600 MBs.

Thanks so much Linc Davis. Hopefully it will stick.

My comment from original article of this tweak/hack about ACPI_SMC_PlatformPlugin.kext :

Actually another thing what deleting this plist results in, is it disables Turbo Boost 2011 and up machines. This isn’t necessarily a bad thing, cause in my case, I have MBA 2011 mid 13inch i5 1.7ghz, it actually reduced the CPU temp under full 100% load from 99C to 84C, fan speed decreased as well from 6.5k to 4.7k. Now the reason for that, is that TDP actually decreased from 18W to 10W, while CPU frequency remains at stable 1.7Ghz at all times. I can’t quite explain the TDP and temp difference, as Turbo boost supposed to work only when one core is fully utilized and the other is pretty much idle. But one theory which comes to my mind is that 18W TDP is actually way more than needed for most chips to run at 1.7Ghz 100% load, meaning that whilst you get that short burst of perfomance (overclock) above the total of 1.7ghz on all cores till you hit 99-100C, this kind of temperature is certainly not doing any good for machine. IMHO TurboBoost is poorly implemented on those machines, but funny enoudh I doubt there is someone to point fingers at, according to specs, Intel’s chip supposed to use 17W running at 1.7Ghz, yet they managed to make one able to run at 1.7ghz with only 10W tdp. Now Apples turbo boost config works according to the Intel specs, it sets the maximum frequency possible to stay at 17W till it hits some 99C, but since it doesn’t take to the account that varying per chip TOTAL frequency (NOT per core/turbo boost) can become higher than stated in specs, it actually overclocks the chip. I believe Apple missed there a great opportunity to take advantage of intel chips and fix heat/noise issues and ironically, save battery power (what this plist file intended to do in the first place).The only “downside” would be that every chip is different, so the temperatures under load would vary abit from model to model, but then again, turbo boost freqs are varying as well, moreover, remember that big story about toshiba vs samsung SSDs in MBA 2011? haha.

Here is a fix to kernel_task issue with high memory and CPU uses

Fixing kernel_task Memory / CPU Problems in Mac OS X

Hope this Helps.

Thankyou.

Hi Linc,

I know this forum might be long dead, but I'm having this issue and it's driving me crazy. Ran your terminal command to see what is overloading the kernal task, and it's: com.ti.driver.TIVCPSerial (1.2.1)

Thought it was CUDA for the longest time (it was the only other thing in the list). I think I finally succeeded in removing that, and this is all that's left apparently. Any idea what the **** it is? Google doesn't give any good results. 😟

Nevermind, I removed the TIVCPSerial.kext.. and confirmed it wasn't running in the kernal.. no 3rd party extensions are, and i'm still getting the kernal overload. My top of the line macbook pro is unusable at this point. And I have 2 videos that need to be edited in 2 days. Fu.......

Boot into Safe mode. Does kernel task still consume 600%. If not, then it is some 3rd party addition you have on your system.

<http://support.apple.com/kb/ht1564>

Suggest you start a new thread.

Post every bit of information Linc has asked for.

Maybe even post the output from EtreCheck (Linc will not look at that, but others might).

<https://discussions.apple.com/docs/DOC-6174>

Continuing to post in this thread is likely to be lost as it is a) long, b) chances are the original post is not the problem you are experiencing, so that just confuses things. Start you own thread and provide as much information as possible.

.

thanks, a permissions repair did the trick as far as the .Kext file is concerned