So it really was a problem of improper subtimings, i.e. RTL values. Thanks, that was concise.

It seems I am a prime example of someone messing with settings I don't really know about and wondering afterwards why it's not stable. I will go and lap the hell out of my chip now and maybe later in the night test again if cas7 is possible with the proper boot up baseclock. Or maybe on auto. ^^;

I hope this 'one boot stable, next boot unstable' isn't coming back.