I'm suggesting 2 killer features that will make this incredible miner complete:
1. Include in watchdog/logfile the uptime of a crashed GPU (i.e. GPU 4 crashed, uptime: 5 hours, 12 minutes)
2. A simple fan curve in addition to the temperature watchdog (i.e. --temp_max=83,100 -> max temp of 83 while fan must be at 100% and --temp_ramp=75,70 -> when it reaches temp of 75, spin fan to 70% and increase to linearly until max defined temp is reached)