Extremely High Wired Memory Usage, used by Kernel_task

[flynnwinch]

So I followed this guide for my most recent build. Everything works pretty well, I simply discontinued the use of whatevergreen because it seemed to cause slow boot and I am running a GTX 1080 on high Sierra (the guide is meant for a Mojave AMD build). However, I have been dealing with huge memory leaks since the machine was up and running. I have 32gb of ram, which gets used up entirely within about 24 hrs, the majority being used by kernel_task (which is currently using 18gb of ram). The only solution is to restart, only to face the same issue the next day. The system becomes unresponsive around 24-25gb total ram usage. If I don't restart it, I'll come to find that my computer is simply off (or sometimes has restarted), consistent with a kernel panic (although I don't get any error message reporting options whatsoever).

I've tried to track down the culprit (specifically this guide), and used a few apps to try to isolate the issue but I've not been able to solve this. The only error that seems to show up consistently this:

Spoiler

Although I've upgraded my hardware from an Asus z170p / i7-6700k to the Designare z390 / i9-9900k, I've not installed any applications that wasn't running previously, and I haven't significantly changed the kexts that are being used, which makes this frustrating. I know that some real Macs running high Sierra also face this issue, but I have not found any solutions that work.

If anyone has any suggestions (or needs me to post any logs for more info), please let me know. I'm getting really frustrated, and I'd love to be able to use my desktop reliably for classes again. My previous build was working flawlessly for 2 years, and I'd really love to be able to bring this up to that standard.

Thank you in advance !

 

[flynnwinch]

After some investigation, i found that I now also have a memory leak in my Windows 10 installation. This seems to show up if I run the CPU at a high clock speed for extended periods of time, such as stress testing.

My hardware is as follows:

-gigabyte designare z390
- i7-9900k
- crucial ballistix sport 32gb (2x16) ddr 4 @ @2666 MHz
- MSI reference gtx 1080
- WiFi + Bluetooth: fenvi t1919 pcie WiFi card BCM94360
- 3x sandisk 3D ssd ( 2x 2tb + 1x 1tb)
- samsung nvme 970 plus 2tb

really hoping for some insight or advice here. I find it strange to have a memory leak show up on both OS after hardware updates, so I really don’t know where to look.

thanks !

 

[flynnwinch]

Anyone have any ideas? :/

 

[pastrychef]

Anyone have any ideas? :/

Did you update the firmware on your Samsung 970 EVO Plus?

The console logs show fsctl errors which are all related to file system.

I don't know if the Samsung 970 EVO Plus has issues with Windows... But it's pretty well known that it has severe problems with macOS. For most people, updating the firmware fixes the problem, but I've seen a few report that even after updating firmware, they still had issues.

 

[flynnwinch]

Thanks for your reply!
I updated the firmware indeed. I was actually unable to boot until the firmware was updated.
I guess I’ll try buying a new drive and testing again. Any recommendations on a drive with comparable speeds compatible with hackintosh?

 

[pastrychef]

Thanks for your reply!
I updated the firmware indeed. I was actually unable to boot until the firmware was updated.
I guess I’ll try buying a new drive and testing again. Any recommendations on a drive with comparable speeds compatible with hackintosh?

Before going out and spending on a new NVMe SSD, you can try using any old drive you may have laying around at home.

In the last year or so, I have been using HP EX920 SSDs for their extreme value and performance. I think the EX900 is also very good.

 

[flynnwinch]

I don’t have another nvme drive. I’ll try booting from CCC clone on a sata ssd and I’ll report back.
Thanks !

 

[pastrychef]

I don’t have another nvme drive. I’ll try booting from CCC clone on a sata ssd and I’ll report back.
Thanks !

Keep in mind that those who had issues with the 970 EVO Plus had issues just with the drive installed. It might be a good idea to pull it out of the system when doing your testing.

 

[flynnwinch]

Keep in mind that those who had issues with the 970 EVO Plus had issues just with the drive installed. It might be a good idea to pull it out of the system when doing your testing.

Thanks ! I experienced that too. Couldn't boot past clover with the drive installed until I updated the firmware.

I'm gonna let it run during the night and see what my kernel_task status is in the morning. It usually takes a few hours anyway

 

[flynnwinch]

Keep in mind that those who had issues with the 970 EVO Plus had issues just with the drive installed. It might be a good idea to pull it out of the system when doing your testing.

So after leaving it on for about 8hrs, booted from a CCC ssd clone without the nvme drive installed, the result is the same. I end up with ~25gb ram used, and kernel_task is using 18gb of that.