So no need to RMA the CPU if I only use Windows (I use it for gaming, music, movies, audio/video rendering, web design)?
Again, this is a hardware problem. If a new version of some program triggers this problem it won't matter what OS you are running.
I run Linux and can reproduce the problem several times per hour by compiling the Linux kernel using all 12 threads. After disabling ASLR I have not been able to trigger the problem again after about 10 hours of the tests that previously caused it.
My advice would be to RMA it if you don't want to live with the risk of being unable to RMA it later.
As for me, I'm in the (presumed) same position as you; I see no glitches, code failures, or crashes. My system is rock solid and nothing fails. However, I know that that might change tomorrow with some new program, release, or just a change in my workload. I don't care. I have spare machines capable of doing everything I need -- and I intend to replace this chip when the first substantial upgrade comes out anyway.
My guess would be that the worst case scenario for you is the very low probility of a short downtime or process interruption and maybe even the price of a new chip. That risk-adjusted price is low enough for me to ignore -- and probably for you too.