Delta IV note?
	"Most Metal Rocket" (Tory Bruno)
	https://www.youtube.com/watch?v=Kg65SiK4-bI
	I couldn't figure out the CPU architecture
		ChatGPT didn't know
		It's not in the 300 page user manual
		Maybe it's a secret...
	How much power could the payload consume?
		https://www.ulalaunch.com/docs/default-source/rockets/delta-iv-user's-guide.pdf
		Page 180 in the .pdf
		11A is per pair

Pipelining:
	Might as well fetch the next instruction while the current one finishes
	We can take that a lot farther
	Hazard (wikipedia has a good example)
	Speculative execution:
		Image you just assume conditional jumps won't happen
		Can now pipeline again
		Being wrong is expensive now!
	Remember, no time to think really
		But we could have a set of transistors...	

Multicore CPUs
	Seems simple:  Just put a bunch of CPU dies together	
	EPYC page:  https://en.wikipedia.org/wiki/Epyc
	Cache:
	L1:  Specific to a core
	L2:  Shared between cores maybe
		ChatGPT thought so
		AMD 9000 series has per-core L2
	L3:  More shared
	So...each thread should have its own registers
		And stack?
		Where are their stacks?
		Can a thread access another thread's stack?
		Do they have the same heap?
			Can a thread allocate something and another thread free it?
		What happens if one thread calls exit?
		Do they really run at the same time at nanosecond precision?
			clock_gettime
	Let's see if we can answer all these questions!