I think this needs to be investigated via the Remote Agent log file on the computers where you run the mining. It's possible to search for "Executing GPU clocking command" to see what commands Awesome Miner sends to MSI Afterburner.
Ok, thanks, I'll check that... at first blush, on each change, I'm only seeing the start commands being executed never the stop.
Shouldn't it do a start before it starts mining and a stop *before* it changes to something else, and then the associated start for the new algo?
Yes, it should. So "Start clocking for first algorithm" -> "Stop clocking for first algorithm" -> "Start clocking for second algorithm" should be the work flow for executing the clocking profiles during profit switching.
If you don't make any progress, you could send me the details via mail and let me know the exact time where it failed - and also what the expected behavior was. Thanks!