I also had a lot of problems with stak-amd... I'm using 8xVega64 and after a while with stak-amd one card goes to 0h/s and then the whole rig just shuts down. I switched yesterday to cast-xmr, it seems more stable and until now I had no crash/shut down. I also noticed if I'm connecting with teamviewer or opening some applications like GPU-Z, the cards are starting to mine with a lower hash rate same like the HBCC was deactivated.
Exactly my issue. One card hangs then the rig follows shortly after.
I also see that any diagnosing utility must be started beforehand or it takes forever to launch and has a high probability of hanging the rig.
Castxmr seems less susceptible to this compared to stak-amd.
Like I said, lowering the speeds of HBM does not help. I go as low as 900 and still hangs. BTW, I'm using sapphire bios. Might be that it's not the same as other makes (gigabyte/msi etc).
One thing I noticed, is stakxmr uses a little more power per card (about 5W).
Running out of ideas to try.