I used the supply margining feature and got the core voltage up to 0.85V. I'm still getting exactly the same results.

I poked around online and found some other test vectors. I don't get the same results as they do either, but I consistently get the same (incorrect) nonce's so I guess that's good I'm running multiple jobs through each chip and looping through the whole test five times (just running one chip at a time currently).

I'm seeing peak to peak ripple of about 50mV on the scope so power looks to be good. I would also expect to not get such consistent behavior if I had a power issue?

So perhaps I have a nasty bug somewhere…

Thanks!