The problem is more that we are still horribly addicted to single core performance and cannot really utilize a proper manycore architecture likes the ones comin

The problem is more that we are still horribly addicted to single core performance and cannot really utilize a proper manycore architecture likes the ones coming out of china's higher end semiconductor industry. This forces the semiconductor industry to keep coming up with clever hacks to make their cores more efficient despite not being able to actually increasing speed by introducing all sorts of "clever logic" that suffers the same kind of bugs as software directly into the silicon, and until we stop doing that spectra like bugs will likely keep resurfacing. At this point the only thing unaffected seams to be low end arm cores(which don't attempt to be smarter then they are), and a couple of obscure legacy architectures like SPARC and Itanium.