It isn't the really slow emulation that most people are used to but dynamic binary translation. Decent ISA translation results in 50-70% of native performance. That's what Itanium chips used starting with Montecito.
Hardware emulation of IA64 in x86 resulted in a 1GHz Itanium 2 performing like a 200MHz Pentium MMX. IA32-EL binary translation of 1.5GHz Montecito resulted in a performance that Intel reported was on par with 1.5GHz Netburst-based Xeons.
Geekbench results show x86 results of the 835 gets ~1200 points, while ARM code gets about 2000. The problem is, Goldmont Plus cores in the Pentium Silver chips have performance on par with SD835's native performance.
But now you are running one chip through binary translation, and one is running native.
There will also be applications that are incompatible, and will put the people using them off using WoA devices once again. Compatibility is hard, there are issues among Windows versions, nevermind binary translation of a completely different ISA!
People think it'll somehow be different this time, but Intel plans to bundle their LTE modems as well, and battery life is roughly competitive between the two. Then the SD835 based devices will essentially become a slow alternative to Intel chips that sometimes have compatibility issues. If Intel stops development of these cores altogether, eventually they'll have a serious encroach of ARM to deal with.
As long as they develop the cores at a brisk pace they are doing now, PC space is out of ARM's reach(just like mobile is out of Intel).
News 
Market Data 
Discover 
AI
