It would really improve my experience with Julia if loading a bunch of big package ecosystems wouldn’t cause a considerable delay while a bunch of stuff gets recompiled. Revise.jl is great but sometimes doesn’t work when for example redefining a struct or whatever. Obviously a faster PC is going to be faster at doing all the stuff that happens when you load a package.
Question is, if you were to build a PC, what do you want to actually optimize for? Is single core performance actually the only relevant metric? Are generic CPU benchmarks representative of this use?
AIUI precompilation can be multithreaded but when you actually load a package, that’s single threaded. I would think things like memory size and GPU are completely irrelevant. Is stuff like memory bandwidth, disk speed, cache size, and parameters like that at all relevant to consider beyond generic CPU benchmark scores?
I kinda like the Apple silicon, and AMD, but last I heard Intel is king of single core performance, so maybe that’s the way to go. A laptop is of course also convenient but if a desktop just has better performance that’s the way to go.