As usual, I spoke too soon
Not long after it failed at 4096K (~9 hours), so maybe a little bump in ram or NB voltage could have helped, but after leaving it running for the day, it failed at 96k on the second run whilst only 3 cores were running. As I'm so close to getting stable, I'm giving it another try with a notch in NB to 1.471v, and 1.89 on the RAM. CPU GTLs were set to 10,-45,10,-45.
0,45,0,45 gave a BSOD on login so I think I could actually be getting somewhere with a 50-55mv gap in the CPU GTLs, but with such a long time between tests it gets really finnicky . And this board is probably just playing silly buggers with me again

Had a missed delivery though, might be my UD3P from Belgium... I see a silver lining