So, I revisted this little experiment.
I was looking over the numbers, and my 3rd chip was running on average ~10°C hotter than the 4th chip before delidding. And yet was performing about the same. Also, when it did crash it was just errors in tests instead of hardlocks and BSODs where the 4th chip I had lots of hardlocks and BSODs during testing.
Thought to myself, maybe it was an esspecially bad TIM application on the die?
So I delidded it! And yup! Where I gained 15-20°C cooler temps on the 4th chip I gained 30-35°C cooler temps on the 3rd chip after delidding. Die was making poor contact with the heatspreader.
The stability tests are showing the results as well. I went straight for full AVX testing this time. 8 hours of OCCT Small.
4th chip (the one I delidded before)
5.0GHz @ 1.33V
5.1GHz @ 1.43V
5.0GHz @ 1.28V
5.1GHz @ 1.35V
5.2GHz @ 1.46V (hottest core hit 86!)
And is easily bench/game stable at 5.3GHz with 1.35-1.38V
Glad I looked at this again. I know for sure which chip I am keeping now.