PC Graphics

[miqome]

This started happening after the AMD Drivers crashed while playing FFXIV. Afterwards the GPU would drop down to 0MHz repeatedly and everything froze, then 2s later the Core Clock started climbing again and the cycle repeated.

I restarted my Computer hoping it would fix the issue, and it definitely got better, but now the Core Clock keeps dropping to 26MHz. I tried re-installing different driver versions that worked before (using DDU+Safe Mode). That gave me another improvement of at least being able to play games, although the Core Clock still seems to drop to 26MHz and at max load the highest it goes is 1400MHz whereas before it would go as high as 1670MHz.

Strangely enough, even if the Core Clock drops to 26MHz there is no change whatsoever in FPS? Tried all kinds of programs to make sure it's not just a software error, but they all claim the same 26MHz (HWInfo, GPU-Z, AMD's internal performance monitoring).

The manufacturer is ASRock, GPU is named Phantom Gaming X Radeon RX VEGA 56. I never overclocked the graphics card at all, or any PC components before for that matter. I have no idea what's going on or how to fix it. I tried googling the issue but only found 1 thread with no response to it from anyone nor update from the author.

If anyone knows what the hell is happening or how to resolve the issue I'd be eternally grateful!

[amdman]

Interestingly enough, while browsing youtube I came across this video with a lot of information I didn't know. I find it astounding that older games sometimes are so easy for these modern cards to render it drops to 500 Mhz core and even lower! I have several old games where they play fine even with the Mhz so low, but on some especially DX 11 games, it can cause problems, and you can set it to not downclock so much. This video will give you some things to try. Please post letting everyone know if it fixes your problems, good luck:

https://www.youtube.com/watch?v=14QSlEvvzoA

[miqome]

Thank you for the response & suggestions! Appreciated. Unfortunately none of that alleviated the issue. Some made it even worse Also, it's not just low clocks while playing old games. In current games it's the same issue. Additionally, somehow, my max clock speed went from almost 1700MHz to 1400MHz and most of the time its 1200/26MHz. I wanna get back the 500MHz of performance. Idk how it just disappeared into thin air

[amdman]

Seems like you have done a lot of good work to try to figure it out. At this point you could make sure your Power Supply is providing ample power, and if you can try the card in another PC to try to isolate the problem. Double check your case isn't causing it to get hot, a last resort would be to repaste and rethermal pad the card. Also you can remove and reinstall the Chipset drivers.

Maybe others have things you can try?

It's a tough one, I hope you find a solution.

[ThunderBeaver]

Did a little research. All I can find is the following.

1. In your Adrenaline Performance to Tuning setting set to custom and set your Power Tuning slide bar to max.

2. Do not lock in any of your framerates also try turning off all sync options as this will also throttle your framerate and cause your clocking frequencies to drop in tandem to conserve power.

3. Final but I find this one dubious at best unless your running hardware that is 6+ years old. You may be experiencing a CPU/RAM bottleneck if this is the case simply check windows update for the latest .Net Framework. Go to your MOBO's website and download the latest CPU Chipset drivers.

4. Finally on your MOBO website check your BIOS update options. A lot of times older rigs will work great if they have an up to date MOBO BIOS.

5. It couldn't hurt to unplug your HDMI/DVI/VGA cable and check the length of the cable for defects (sharp bends that may be an indicator of broken shielding and conductor cracking/dry rot maybe a pesky mouse got to your cables. DVI/VGA cables the most common issues is broken missing pins on both ends or sometimes just need a good cleaning.

6. Try blowing the dust out of your tower. Even a thin layer can cause a host of issues. Make sure to power down and unplug power when doing this to reduce risk of static discharge. Take your GPU out and inspect the heatsink and fan for dust buildup if you find clean it out.

7. If you have a smart monitor/tv for your display connect it to the interment and update its firmware.  

[amdman]

A couple of more thoughts:

Check the temperatures during intense games or benchmarks, is it hot?

Use GPU-Z to check that the MHz listed for the card is read by it, and it is the same as your manufacturer's specifications.

Check for basic problems like Windows corruption with sfc /scannow.

Make sure you are running a separate power cable to each power input on your GPU.

[amdman]

Maybe this could help:

"Check your windows performance settings and your AMD Adrenaline performance settings."

https://community.amd.com/t5/graphics/my-rx-6600-is-running-at-20-mhz-clock/m-p/509587#M82839

[miqome]

Thank you very much for the long, well-thought out reply. Very much appreciated! I'll go 1 by 1.

1. The graphics card doesn't seem to react to AMD tuning settings anymore. No matter what I set it to, it's always the same result I described above or an outright complete crash of the system.
2. I do have my frame rates lock, however I also tried different benchmarking tools including AMDs own internal tool and it doesn't matter. Card won't clock as high as it should and keeps dipping to 26MHz.
3. My CPU is a 5600X and my RAM is G.SKILL Ripjaws V 4000MHz (Running at 3866MHz due to stability issues at 4000). So this shouldn't be the issue and never has been before the strange crash that caused the problems to manifest.
4. I did that, including latest AMD chipset drivers, but unfortunately it made no difference whatsoever either.
5. I have tried 3 different cables and monitors of various ages, resolutions and framerates. Unfortunately, no difference.
6. I went ahead and cleaned out my whole computer including disassembling CPU + applying new thermal paste. I don't trust myself enough to remove the plate of the GPU and do the same though. Temperatures of the GPU max out at 80-85°C under full load so should be good too.
7. Just using regular displays so that option is unavailable to me.

I think I did everything in power to fix it including asking the all-mighty interwebs, so I'm just gonna contact the vendor I got it from now and ask them to check it out and if they can't fix it to replace it. I'm still open to new ideas and will check back here until they get back to me and, hopefully, take GPU of my hands and replace it. Will update once that's the case so nobody wastes their time to keep try and help me. 

And once again, thanks to everyone for their suggestions and the time-invested trying to help me. I very much appreciate it and hopefully someday, someone with a similar issue may find this thread and it'll help them out even if it didn't help me yet

[amdman]

"Temperatures of the GPU max out at 80-85°C under full load so should be good too."

This would lead to ask what is the GPU Hotspot Temperature getting to? 80-85 GPU Core is quite hot and probably the GPU Hotspot Temp is going too high, here is a reddit post that sounds very similar to your problem and may help others in the future:

https://www.reddit.com/r/Amd/comments/bcyukn/vega_56_early_thermal_throttlingplease_help/

I see you are planning to get a replacement, it's possible there is a fan or thermal paste or thermal pad problem with the card. I hope you get a good card to replace it.

[miqome · Accepted solution]

I believe that I found the solution to my problem, I haven't done any extensive testing yet but this reddit post outlined a way to export the BIOS and doing something with regedit. After this the voltage seems to be able to climb again to advertised values, same with the core clock + no more weird random 26MHz readings from monitoring software.

I feel so stupid for not checking the voltage when this problem first manifested, wasted so much effort when I could've just scrolled up in HWInfo64 ;_;

[amdman]

That's why it's always a good idea to consider not overclocking. I worked with a person that encouraged people to overclock, and when the people that listened to Bobby would experience problems, destroyed hardware, the people would say "but Bobby said it wouldn't be a problem and is safe!" Turns out some people will lie right to your face. Why do the companies have stock tested values for hardware? Because pushing hardware past what it was designed for is dangerous. Overclocking will wear your hardware faster for a small increase in performance. But everyone thinks it's so cool. Kind of stupid if you ask me.