No HBCC toggle.

HBM2 clocks drop way too aggressively for low GPU utilisation / idle (down to 167Mhz), such that if you have hardware acceleration enabled in browser (tested on Chrome and Opera), the PC will become completely unresponsive for 2-3 seconds as any Flash or HTML5 (or other) video loads in the browser.  The clock will jump from 167Mhz to max clock after about 2 seconds and the computer will then unfreeze and the video will load, HBM2 clock then drops again to idle.   Others have reported this too.  I can see why you'd aggressively drop clocks like that for enterprise cards, but consumer / gaming? 167Mhz seems like overkill.   Anyway, have had to disable hardware acceleration.   There are occasional other freezes, like loading Skype GUI from idle state, or other similar scenarios.

No individual game profiles apply any of the settings therein.

The voltage settings in WattMan are broken / very mixed up.  Supposedly you can set HBM voltage (no you can't).  GPU P6 and P7 state undervolting do not apply properly unless you apply an identical value in the HBM voltage parameter.  This makes no sense, and a few people, me included, had to work this out through trial and error.   It allows you to set between 800mV and 1200mV, but won't apply anything below 950mV.  On the plus side, my card is rock solid at 1630Mhz under load at 950mV (confirmed via HWiNFO64 measurements), with 1100Mhz HBM2.  On air.  Making it draw similar / less power than a 1070 for significantly better than 1080 performance.

I'm not 100% sure if it's HWiNFO or RTG driver team's fault, but I'm assuming the mislabeling is at a driver level, as HWiNFO just usually spits out all the sensor readings.  GPU Core Voltage and GPU Memory Voltages are the wrong way round - "GPU Core Voltage" is locked solid at 1.356V as you'd expect for HBM2.  "GPU Memory Voltage" is dynamic with GPU load and maxes out around the value you set for P6 & P7 (albeit also having to set so-called HBM Voltage - what ever that value is - in WattMan).  Also, you should really expose readouts in WattMan ...

If MSI Afterburner is opened even once, then it scrambles Radeon Settings, and factory reset, uninstalling Afterburner or restarting the PC have no effect.  It remains borked and you have to uninstall drivers, use AMD Cleaner tool, then (in my case) CCleaner registry cleaner, then reinstall.  Should you open it, it does the following - HBM overclocking becomes a jumbled mess of numbers in Wattman (-52900Mhz and always reverts to this as soon as you press Enter), Power Limit can be set between -1%, 0% and +1%.

In PUBG (PU's Battlegrounds), there appears to be heavy over tesselation going on.  This isn't observed on Fiji or Hawaii or Polaris (or NVIDIA) cards I have access to.  Player model shows very heavily 'goosebumped' skin and some rocks have very heavy surface deformation.  Maybe game profiles not applying has somehow defaulted it to max or very high value tesselation? If not, NVIDIA partnered game funny business?

Radeon settings forgets custom settings if you change to Turbo / Power Saver / Balanced (though why you would given the NEED for undervolting I'm not sure).  Also, why no saveable profiles?

Fan profiles for the 3 stock profiles are way too low, IMO.  Thermal throttling kicks in far too quickly, and given the sound profile of the fan (mostly blowing noise and no rattle or whine), it would do to increase RPM quite significantly at least on balanced / turbo.

There's more, but those are the main things that stick out so far.