Patrick from STH gives his opinion about where AArch64, AMD, and Intel are going. I found his views interesting on how each different company sees the future.
A good summary… but I would have said…
Intel = $$$ for accelerators that do not always get used, and expensive and slow without them
AMD = Faster than Intel when no Intel only accelerators are used, cheaper than Intel .
ARM = Overall best bang for the buck, Way faster than Intel when there are no accelerators, still faster than AMD when comparing vPCU . Needs faster memory access. Maybe some accelerators down the road.
I would add in better performance per watt for ARM64
I’ve benchmarked use cases like 8 cores of ARM vs 8 threads of AMD… ARM blows it away when accellators are not used.