Contribute
Register

Vega 64 Random Black Screen + 100% Fan Speed Issue

Status
Not open for further replies.
Joined
Jan 11, 2021
Messages
8
Motherboard
Gigabyte Z390 Designare
CPU
Intel i9 9900K
Graphics
Sapphire Nitro+ Vega 64
I am running into issues with my build recently.

Basically, it appears that under high GPU load activities, such as video editing, gaming, and 3D rendering, my computer will randomly go to black screen, and the fans will instantly ramp up to 100% speed. Note that if audio is playing, it will continue to play normally until the computer is turned off.

- My first theory was that the card was overheating. My fans never spin, even under high load.

- My second theory is that because the card is supporting two 4k monitors, that any extra load ontop of that can cause the card to become unstable.

- My third theory is that the card itself is just defective

- The Fourth theory is that the card is having issues with power, however, this one is at the bottom of the list because I purchased a power supply with enough wattage for my whole system.

So far I have tried using VEGATAB for controlling the fans, but that didn't seem to do anything. Also, when I unplug one of the 4K monitors, the likelihood of random crashes reduces in frequency greatly.

I am running Mac OSX 10.14.5 Mojave with Clover install

I am wondering if anyone else has had this issue. I have seen similar things but nobody seems to have had the exact same problem as me. It is starting to get frustrating because I rely on this computer for video editing and it has made the process 10x more time consuming because of these random crashes. If anyone can help me get to the bottom of this it would be greatly appreciated.
Thanks
 
Your graphics are crashing for some reason. I had this issue when I lived in New Mexico and my room was carpeted. a static shock would cause exactly what you are describing. if you aren't having a static problem, my first question would be to ask what your power supply is. Low power could cause your graphics card to crash. Neither your Vega or your i9 9900K are power sippers. My second question is what is keeping you on Mojave? Vega drivers weren't very good during Mojave but most of the issue have been solved. I don't know if the improved drivers were backported to Mojave.

I think you are dealing with a power delivery issue. if possible, see if you can recreate the issue in Windows.
 
Your graphics are crashing for some reason. I had this issue when I lived in New Mexico and my room was carpeted. a static shock would cause exactly what you are describing. if you aren't having a static problem, my first question would be to ask what your power supply is. Low power could cause your graphics card to crash. Neither your Vega or your i9 9900K are power sippers. My second question is what is keeping you on Mojave? Vega drivers weren't very good during Mojave but most of the issue have been solved. I don't know if the improved drivers were backported to Mojave.

I think you are dealing with a power delivery issue. if possible, see if you can recreate the issue in Windows.
First off, thank you for the reply, I really appreciate it.

Here is a link to the power supply that I have https://www.amazon.com/gp/product/B005BE058W/?tag=tonymacx86com-20

I do have carpet on the flooring, so it could be a static shock problem. How would I go about checking/fixing if that is the case?

I have so far stayed on Mojave because the thought of upgrading to a newer OS seems very challenging. I am decently tech savvy, but not the most advanced at it. The possibilty of software not functioning on the newest OS is my biggest concern.

Also, I make music and run alot of VST plugins on my computer that may or may not be cracked. So updating those all to the newest versions is nearly impossible for me to do. So I would basically risk not being able to open any of my projects if some of these plugins were to break.

I was thinking about doing a windows 10 dual boot on my machine for gaming and just to see if the same problem would happen to me in that environment, just have to order and SSD for that, so maybe I will be testing that in the future.
 
Your power supply is adequate and of good quality.

If you were dealing with a static issue, the issue would happen when you touched something (you know what a static shock feels like). I would strike that out as well.

This really seems like a hardware issue to me. If I were you, id try to get into a Windows, or even linux environment and do some stress testing to try to recreate the issue.

one last thing, if you were messing around with vegatab or any other powerplay table modification, you should remove those. You might be unknowingly overclocking or undervolting something while trying to modify fan speeds.

The fan speed issue is something that has been solved with OS updates and shouldn't be an issue anymore.

sorry I'm not able to help any more. I guess my suggestion is to try to take macOS out of the equation. You might be able to make a live Ubuntu install USB and run some kind of GPU benchmark from the USB drive.
 
Your power supply is adequate and of good quality.

If you were dealing with a static issue, the issue would happen when you touched something (you know what a static shock feels like). I would strike that out as well.

This really seems like a hardware issue to me. If I were you, id try to get into a Windows, or even linux environment and do some stress testing to try to recreate the issue.

one last thing, if you were messing around with vegatab or any other powerplay table modification, you should remove those. You might be unknowingly overclocking or undervolting something while trying to modify fan speeds.

The fan speed issue is something that has been solved with OS updates and shouldn't be an issue anymore.

sorry I'm not able to help any more. I guess my suggestion is to try to take macOS out of the equation. You might be able to make a live Ubuntu install USB and run some kind of GPU benchmark from the USB drive.
I would have to agree with you. I removed Vegatab awhile ago, because it was causing some other issues as well. As far as going about upgrading to 10.15, has there been a reliable way to update released yet?

I will plan on getting a new SSD soon to install windows onto, and see if I can run some sort of diagnostic tools. If you or anyone has recommendations on where to go for those, feel free to drop them below.
 
Hey Matt, I'm in almost the exact same boat as you. I have the 10900, but also the Vega 64. I have tried SO MANY things.. and I'm on Catalina with Opencore. So I hate to tell you that upgrading might not solve your problem, cause it sure hasn't solved mine!! I just built my machine in September. My power supply is 50 more watts than yours. So maybe we are both having hardware issues? Wondering if you've had any success since your last post.
 
Hey Matt, I'm in almost the exact same boat as you. I have the 10900, but also the Vega 64. I have tried SO MANY things.. and I'm on Catalina with Opencore. So I hate to tell you that upgrading might not solve your problem, cause it sure hasn't solved mine!! I just built my machine in September. My power supply is 50 more watts than yours. So maybe we are both having hardware issues? Wondering if you've had any success since your last post.
So far the only thing that seems to have worked for me is using only 1 monitor when doing GPU intensive tasks, such as video editing. Having both 4K monitors running seems to overload the GPU memory and seems to be what is causing the crash. I have a hunch that it might not be a hardware issue, but something that is just wrong with OSX's integration with the Vega 64 Driver Memory.

For example, the GPU memory at idle is generally somewhere around 20-30%, under load maybe 60-70% (with one monitor). After plugging in the second 4K monitor, the memory usage jumps up to 90-99% range, which would then cause any other applications to be stuck or cause overload. I think this is just poor poor optimization in the drivers and I don't know if there would really be a good fix for it other than replacing the graphics card with something else.

I have been meaning to get a new SSD and try a Windows 10 installation just to test if the same problems would happen on Windows, but haven't gotten to it yet,.
 
Here's a thought, plug in two separate 8 pin power cables to you GPU, I was experiencing the exact same things you were describing above, but on an older haswell build, with a Vega 56, running catalina.

Then I saw Morgonaught's video about plugging in two separate 8 pin power cables on her X 6800 setup, instead of just using the one cable with two heads plugs on one end. So I went ahead and tried that, and I'm currently running some transcodes in resolve 17, which were previously causing the freezing crashing type behaviour you described above and so far so good, stability has really improved, the vram on the GPU isn't maxing out anymore and temps under load have gone down from 60degrees to 80 degrees to crash, to sticking at about 50 to 53 degrees. I hope that helps.

--- Update---

so the 8 pin connector helped for sure but I was still getting crashes under heavy rendering loads, the fans would not spin up properly soon or fast enough, and peak in temperature would cause a crash.
so a little more research led to...


it turns out AMD GPU's are natively supported, but,,, you still need do a couple of tweaks, it's in the above guide

max temps under full load have now dropped down to a nice 45 degrees and idle is in the 30's .

many thanks to @ Edhawk who's post led me to the guide.
 
Last edited:
Here's a thought, plug in two separate 8 pin power cables to you GPU, I was experiencing the exact same things you were describing above, but on an older haswell build, with a Vega 56, running catalina.

Then I saw Morgonaught's video about plugging in two separate 8 pin power cables on her X 6800 setup, instead of just using the one cable with two heads plugs on one end. So I went ahead and tried that, and I'm currently running some transcodes in resolve 17, which were previously causing the freezing crashing type behaviour you described above and so far so good, stability has really improved, the vram on the GPU isn't maxing out anymore and temps under load have gone down from 60degrees to 80 degrees to crash, to sticking at about 50 to 53 degrees. I hope that helps.

--- Update---

so the 8 pin connector helped for sure but I was still getting crashes under heavy rendering loads, the fans would not spin up properly soon or fast enough, and peak in temperature would cause a crash.
so a little more research led to...


it turns out AMD GPU's are natively supported, but,,, you still need do a couple of tweaks, it's in the above guide

max temps under full load have now dropped down to a nice 45 degrees and idle is in the 30's .

many thanks to @ Edhawk who's post led me to the guide.
I'm going to order the cables for trying out the dual 8-pin cable setup. I'm a bit confused about the Radeon Compatibility Guide you shared a link to.

It says in the guide that the Vega 64 series needs "Needs RadeonDeInit, early drivers". I just am not having an easy time comprehending what that means. Is the RadeonDeInit the piece of code that's listed below the model guide? And by early drivers, does that mean I need to get and install those from the Radeon website?

I read the guide but I think it's a bit out of my leauge. Would you mind helping/sharing what you had to do to get those fixes working?
 
No what the guide meant be 'Early Drivers' is that at the time of writing the guide the AMD drivers for the Vega series were only recently released by Apple. That is no longer the case unless you are running any old version of macOS. Sierra or earlier is what I mean by old version of macOS.

The 'Radeon Deinit' fix was made part of Clover and could be added via the Graphics section of the Clover config.plist, as shown in the screenshot from Clover Configurator below:

Screenshot 2021-10-12 at 16.24.32.png

RadeonDeinit option is highlighted on the screenshot above.

Alternatively you can use a custom 'SSDT-Radeon-Deinit.aml' if you are using OpenCore as your boot loader in place of Clover.
 
Status
Not open for further replies.
Back
Top