Uhh I filed a bug here and on AMD on Aug 9th with 0 responses, but this is the same issue.
The AMD bug report is much more thorough. I’ve have pretty extensive Linux knowledge and I haven’t been able to pinpoint the problem. At the moment I’m going through trying different suspend modes, i.e. suspending to memory instead of disk or perhaps a hibernate that completely re initializes the GPU as if it were rebooting.