Contribute
Register

High Sierra Random Freezing

Status
Not open for further replies.
Bros, have you tried a tool named memtester (which can be obtained by port or brew).
On my side, my memory always get mistake about 88098809. And it seems that my random freeze is related to such mistake. However, I've tried to run memtest from boot and memtester in linux with very long time with no error reported. So I think such memory mistake should related to Mac OS X.

But I have no idea about the next step.
 

Attachments

  • 6EBB2D24F43A9B9B79B34F3488FAFFB5.jpg
    6EBB2D24F43A9B9B79B34F3488FAFFB5.jpg
    475.4 KB · Views: 181
I'm back on Sierra for the time being as I was unable to solve the Freezing issue.
In my case - the GUI just stopped responding, but I could ssh in from another mac and issue commands, run cli programs etc.
I couldn't reboot or shutdown remotely though - my session would log out from the remote computer, but the frozen Hackintosh would just sit there and not reboot.
I could replicate the issue by leaving my computer idle for a few minutes and then launch a web browser - instant freeze.
I ran geekbench, cinebench and upgraded all my outdated macports in terminal all while TimeMachine was performing a back-up and spotlight with reindexing and the hackintosh handled it fine.
Launch safari or chrome - instant freeze.
Log in remotely (ssh), issue reboot - hackintosh just sits there frozen and won't reboot.

I'm using Radeon GPU (R380X) and Skylake CPU (6700K) on Z170X board.
Sierra is 100% stable and still supports some old 32 bit audio programs that I need to run, so I guess I'm staying put.

Just wanted to chime in to let people know it happens with ATI GPU - it's not an NVidia only issue.

I recently experienced the exact same symptoms - windows frozen, key strokes don't work, mouse clicks don't work - but mouse movement does. SSH from another mac can run top etc. Can execute 'shutdown' but mac screen stays frozen - now mouse does not move.

In my case it was a 10.13.4 updated system with a new ASUS ROG STRIX Radeon RX580. Switching the GPU to a Nvidia 1070 and problems went away. One other thing I noticed, seemed to be related to a specific user account on my Mac. If I logged in with my backup account there were no problems at all, even added a 3rd user and that also had no problems. The freezing only happened with my main user account and did so straight after I logged in, while the accounts auto-start/login items programs were loading. If I disabled my account Login Items (using Shift key at password prompt) then I see a lag, stuttering of windows opening before a total freeze.

So, for me at least, seems it's AMD GPU and related to some app(s) starting at login. Changing to another user account or replacing the GPU with NVIDIA fixed my problem - system now stable.
 
I'm happy to report 5+ days of uptime with no issues.

After months of debugging, I finally found that the issue was related to a combination of two factors: buggy Nvidia drivers and bad AVX offset (I detail this matter in a previous post on this thread).

Any old Nvidia drivers prior to the latest 387.10.10.10.30.106 were not correctly supporting my Nvidia Titan X Maxwell:

As a result, every 2-4 days the driver would randomly panic, leading to the CPU stuck in a loop with AVX instructions and the screen frozen. And because my AVX offset was incorrectly set that loop was triggering a Kernel Panic before the Nvidia Driver issue could generate a bug report to the system logs. Visually it was seen as a hard freeze with no auto reboot, and no SSH connection possible to the machine. The only system log available was the Kernel Panic that would state that something happened (cf my previous posts)…

After adjusting the AVX offset to a proper value, I would still get these random freezes (which were no hard freezes, as the system would be able to recover), the Kernel would be able to restart the Nvidia Drivers. Visually, what would happen is that for ~20 sec there would be a high CPU load just like before. After ~20 sec the screen would turn black and restart WindowServer, so bringing me back to the login screen and effectively killing my session including all running apps. Also the Nvidia drivers would generate a bug report in the system logs! Finally, giving me a clue about what was going on.

When the 387.10.10.10.30.106 got released, big relief… no more random crashes. The issue was gone, and I can now fully enjoy my hackintosh.

What helped me debug this issue was Fortnite. This game was somehow able to trigger the exact same random freeze that I was getting at random occasions (every 2-4 days of uptime). This is what helped diagnose the issue faster and how I discovered the AVX offsets were part of the equation! The latest Nvidia drivers are still a little bit buggy (i.e. getting a few crashes from time to time with Fortnite), but at least I don't get random freezes.

For those out there who try to debug their issues, the best is to make sure not to have any Overclocking-enabled and make sure that your system is stable on Windows and Linux. I could have avoided months of debug if I had left all my OC settings as default (with the default AVX offset settings)... Although that wouldn’t have solved the buggy Nvidia drivers but that would have given me a clue about what was really going on.

Good luck!
 
I'm happy to report 5+ days of uptime with no issues.

After months of debugging, I finally found that the issue was related to a combination of two factors: buggy Nvidia drivers and bad AVX offset (I detail this matter in a previous post on this thread).

Any old Nvidia drivers prior to the latest 387.10.10.10.30.106 were not correctly supporting my Nvidia Titan X Maxwell:

As a result, every 2-4 days the driver would randomly panic, leading to the CPU stuck in a loop with AVX instructions and the screen frozen. And because my AVX offset was incorrectly set that loop was triggering a Kernel Panic before the Nvidia Driver issue could generate a bug report to the system logs. Visually it was seen as a hard freeze with no auto reboot, and no SSH connection possible to the machine. The only system log available was the Kernel Panic that would state that something happened (cf my previous posts)…

After adjusting the AVX offset to a proper value, I would still get these random freezes (which were no hard freezes, as the system would be able to recover), the Kernel would be able to restart the Nvidia Drivers. Visually, what would happen is that for ~20 sec there would be a high CPU load just like before. After ~20 sec the screen would turn black and restart WindowServer, so bringing me back to the login screen and effectively killing my session including all running apps. Also the Nvidia drivers would generate a bug report in the system logs! Finally, giving me a clue about what was going on.

When the 387.10.10.10.30.106 got released, big relief… no more random crashes. The issue was gone, and I can now fully enjoy my hackintosh.

What helped me debug this issue was Fortnite. This game was somehow able to trigger the exact same random freeze that I was getting at random occasions (every 2-4 days of uptime). This is what helped diagnose the issue faster and how I discovered the AVX offsets were part of the equation! The latest Nvidia drivers are still a little bit buggy (i.e. getting a few crashes from time to time with Fortnite), but at least I don't get random freezes.

For those out there who try to debug their issues, the best is to make sure not to have any Overclocking-enabled and make sure that your system is stable on Windows and Linux. I could have avoided months of debug if I had left all my OC settings as default (with the default AVX offset settings)... Although that wouldn’t have solved the buggy Nvidia drivers but that would have given me a clue about what was really going on.

Good luck!
Thanks for the info. Do you imply that you have updated your os to 10.13.4 (for 106 driver)?
 
Thanks for the info. Do you imply that you have updated your os to 10.13.4 (for 106 driver)?

Yes. I've been on 10.13.4 for quite a while, way before these drivers got released.
 
Yes. I've been on 10.13.4 for quite a while, way before these drivers got released.

So if I set the AVX offset to a very high value (say 10), and if my freezes are of the same reason to you, the freeze will be replaced by GUI reset?
 
So if I set the AVX offset to a very high value (say 10), and if my freezes are of the same reason to you, the freeze will be replaced by GUI reset?

Assuming your issue is the same as mine, this is indeed correct. Best way to test is to install Fortnite and check if you can play the game 2-3 rounds in a row (not that I'm a gamer... just that there is a lack of GPU stress test on MacOS and I don't know any other recent app that utilises Metal intensively). That's how I was able to trigger the issue.
 
I'm happy to report 5+ days of uptime with no issues.

After months of debugging, I finally found that the issue was related to a combination of two factors: buggy Nvidia drivers and bad AVX offset (I detail this matter in a previous post on this thread).

Any old Nvidia drivers prior to the latest 387.10.10.10.30.106 were not correctly supporting my Nvidia Titan X Maxwell:

As a result, every 2-4 days the driver would randomly panic, leading to the CPU stuck in a loop with AVX instructions and the screen frozen. And because my AVX offset was incorrectly set that loop was triggering a Kernel Panic before the Nvidia Driver issue could generate a bug report to the system logs. Visually it was seen as a hard freeze with no auto reboot, and no SSH connection possible to the machine. The only system log available was the Kernel Panic that would state that something happened (cf my previous posts)…

After adjusting the AVX offset to a proper value, I would still get these random freezes (which were no hard freezes, as the system would be able to recover), the Kernel would be able to restart the Nvidia Drivers. Visually, what would happen is that for ~20 sec there would be a high CPU load just like before. After ~20 sec the screen would turn black and restart WindowServer, so bringing me back to the login screen and effectively killing my session including all running apps. Also the Nvidia drivers would generate a bug report in the system logs! Finally, giving me a clue about what was going on.

When the 387.10.10.10.30.106 got released, big relief… no more random crashes. The issue was gone, and I can now fully enjoy my hackintosh.

What helped me debug this issue was Fortnite. This game was somehow able to trigger the exact same random freeze that I was getting at random occasions (every 2-4 days of uptime). This is what helped diagnose the issue faster and how I discovered the AVX offsets were part of the equation! The latest Nvidia drivers are still a little bit buggy (i.e. getting a few crashes from time to time with Fortnite), but at least I don't get random freezes.

For those out there who try to debug their issues, the best is to make sure not to have any Overclocking-enabled and make sure that your system is stable on Windows and Linux. I could have avoided months of debug if I had left all my OC settings as default (with the default AVX offset settings)... Although that wouldn’t have solved the buggy Nvidia drivers but that would have given me a clue about what was really going on.

Good luck!

What you detailed sounds exactly like what was going on over here only I have ATI cards. I was using a Gigabyte Vega and was giving me crashes as you describe, I blamed the drivers others told me it is a bad card. I do not really know what the problem is exactly, but WoW + iTunes = Crash (not always with errors) with the Gigabyte card. Now that I am using MSI Vega 56 I have only had one crash in 4 days and I could have triggered that crash on a non-hack Mac!
 
Assuming your issue is the same as mine, this is indeed correct. Best way to test is to install Fortnite and check if you can play the game 2-3 rounds in a row (not that I'm a gamer... just that there is a lack of GPU stress test on MacOS and I don't know any other recent app that utilises Metal intensively). That's how I was able to trigger the issue.
Will you please share your EFI?
 
Assuming your issue is the same as mine, this is indeed correct. Best way to test is to install Fortnite and check if you can play the game 2-3 rounds in a row (not that I'm a gamer... just that there is a lack of GPU stress test on MacOS and I don't know any other recent app that utilises Metal intensively). That's how I was able to trigger the issue.
Unfortunately, I just try to update my os to 10.13.4 and the corresponding Nvidia driver. The freeze keeps apprearing...
 
Status
Not open for further replies.
Back
Top