Bearing in mind we don't know where the temperature probes are reading from for the GPU, it can be hard to figure out exactly whether it's getting to silly temperatures. i.e. if the temperature you're reading is from the sink but it's poorly fitted, not getting the heat away from the core and then the throttling mechanism is using an on-die thermistor then it's possible you're throttling due to core overtemp but not seeing the same temperature within the software.
Have you tried popping the old card back in? I ask as it's always possible you've knocked something during the upgrade process and it's sensible to see if the situation resolves with the old card. You need to isolate the card as being the source of the problem and you can either use the old card in the same system or better yet pop the new card into a different system and see if the problem repeats. I wouldn't bother trying to do this with integrated graphics as there just won't be the poke there to ensure other components are stressed.
We did have someone with something similar recently and I think they reinstalled Windows of all things and it fixed it. Obviously with Windows 10 this is less of a mammoth undertaking than it used to be but I'd check out the hardware first.