80c is at the upper limits of safe for stress testing. A couple of suggestions, try manually adjusting the bios rather than using the auto oc. Auto overclocking or 'easy' overclocking options have been known to be too aggressive and in an attempt to get a more stable overclock sometimes use higher vcore voltages than necessary and that will drive temperatures up.
I'm not sure what version of prime95 you used, if you used the latest version try uninstalling it and installing v26.6 (slightly older version, more compatible with haswell and haswell refresh cpu's) and try using small fft's to get a stable load. This should result in lower stress temps if you were using the newest prime95 version.
This explains it a bit better:
http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html
That's not a bad cooler but keep in mind that even though it's a water cooler it probably won't handle load temps any better than a good air cooler and won't have the performance of a dual or triple radiator cooler.
This may give you some useful info on overclocking:
https://www.youtube.com/watch?v=QBfXruwe8w4&feature=youtu.be&a
http://www.overclock.net/t/1411077/haswell-overclocking-guide-with-statistics
A good idea is to have monitoring software installed and keep an eye on things like vcore (voltage to processor) and cpu core temps. Also if running a stress test program like prime95, occt, intel burn test etc my advise is to watch it with temp monitoring open. Don't just set it up and go do something else. While it's not all that common for steady temps to just spike for no reason, it is possible and if that happens you need to be able to shut down the stress test program. Modern cpu's are supposed to have thermal throttling and safety limits in place to help prevent damage but it could be risky to rely on them completely. Better to be too cautious than have an unusual occurrence that leaves you with a dead cpu.