1. I use same code for all my recent miners: ZEC, ETH, CN. So this problem is not related to this miner only.
2. This problem is related to buggy drivers. Miner sets fan speed but drivers sets some different speed. Then miner reads current fan speed and drivers report wrong speed. Therefore miner calculates wrong new speed, drivers set it wrong again, it increases error in speeds even more and so on.
As far as I remember, I saw at least two reports that this problem is solved in some drivers version. Need to find these messages. Also I will check it on latest drivers, if AMD still has this problem may be I can find some new workaround.
You can try to set static fan speed, for example, "-tt -50". And see what speed was set.
1. I didn't have tested all of them, but I have the same problem with your EthMiner.
2. In the log it's showing GPU0 and GPU6 and the stats are showing GPU0 und GPU1. I've read about problems with the enumeration of the devices in the faq. Do I need to set the value for the 1st and 7th GPU (-tt 75,00,00,00,00,00,70) or only for the 1st and 2nd (-tt 75,70) and you're mapping it internally from 2 to 6?
I've tried to set -tt -50 but there was no effect (it's cooler because I've started it with -li 1):
GPU0 t=66C fan=46%, GPU1 t=65C fan=27%
I'm using the 16.12.2.2-Jan3 driver.
I've also tested sgminer with "--temp-target 77 --auto-fan" with the same machine,card & driver and there it's working fine after some "learning"-time:
GPU 0: 74.0C 2449RPM
GPU 1: 74.0C 1308RPM