I tried DDU to clean install the driver and the phenomenon is not related to any running app. A clean reboot with only the essential drivers loaded will still have the same issue with driver after 184.108.40.206.
In fact I want to down clock the VRAM instead of OC it. The issue here is since the OC/Downclock doesn't works but the core voltage is still dropping, the VRAM can not maintain stable at desktop usage.
I have a 7970 oc'd on win 8.1, 16.6.1 driver.
Had no problems with clocks sticking/freezing/downclocking since 16.3.? drivers.
I usually run GPU-Z render test (offline/online) to check for clock problems/stability with each new install since the earlier crimson problems.
It's definitely a driver issue though. My card sits at 0% activity, all that's on is my desktop, and it locks the clocks in high. I'm having the manually run the fan because 60 degrees idle is a disgrace. It's only due to the driver balls up. Sadly not enough people are suffering from this issue so it's getting swept under the carpet and those of us with it just have to deal with having a burning card.
My card is constantly sat at this with the latest drivers. I had the fan on 50% earlier to cool it down but set it to defaults for the purpose of this screen shot. It's now 61 degrees.
As you can see, 16.6.1.
I even tried altering the clocks with MSi Afterburner but it just reverts everything straight back to 501/1250 on my card. Definitely the driver and not the software. Sadly I need the most up to date drivers to play some games so I'm stuck until I swap to Nvidia.
I can only suggest then, that you send an issue report to amd.
As you know, this is mostly a user to user forum (with some input from a few amd techs).
I believe ray_m has some involvement with reports.
This has been an ongoing problem ever since Crimson was released! I've just downgraded to the last 'stable' build I have an install file for '16.2.1'
I made a thread a while back about the clocks freezing on. Never got a proper answer so I sat on 15.11.1 beta for a while. that was the last true stable driver. It was one of the last CCC drivers.
It's weird but I'll note that my core clock with the recent drivers has been freezing at 501mhz too with the memory clock freezing at the highest it was last set. Previously it had BOTH clock speeds stuck on full. I have a 7950 3GB so for me it stuck on 850mhz and 1450mhz respectively and kicked some wild heat out!
I will also note that I've got crimson 16.2.1 up in the background to check the clocks and they're sat happily at 300mhz and 150mhz with a cool 38-42 degree idle (well, cool for this card).
Like I said, it's been a problem since crimson was spewed out to us and AMD seems reluctant to resolve this one. The one time they did has since been reverted. It's total BS. I wasn't getting the corruption that you're getting though, just stuck clocks. and 60 degree heat constantly like it was under load. I'm shocked it's not fried yet. When it was sticking both clocks full it'd be idling at 80 degrees plus with no way to reduce the clock speeds.
Come on AMD, sort it out. You've already pushed me over to intel, I'll also be with Nvidia soon if this doesn't change.
Just had a youtube vid on and the clocks went to 501mhz and 1250mhz (no overclock on the memory) and as soon as I closed the tab the clocks returned to 300 and 150 straight away. This didn't happen for me on 16.6.1. They just froze.