I ask this not as a gotcha (I don't know the first thing about this), but rather because I'm interested: How do you know not "ever"?

Like, trivially, it's an ASIC, so I can use it to simulate a von Neumann[*] machine, hence I can use it to run whatever algorithm I want. Would that be more efficient than using a modern OoO superscalar? Almost surely not, but that doesn't mean it can't be done, just that it shouldn't be done that way.

*: I realize that the ASICs used in Bitcoin miners don't have dram access, but that isn't a general limitation of ASICs, just those ASIC 'chips' (and maybe not even those chips, just their implementations in bitcoin miners)

EDIT: Thanks to everyone who answered! For some reason, I had it in my head that the way we implement fixed function stuff in an ASIC was basically the same as a "burn once" FPGA. Brains gonna brain.

>Like, trivially, it's an ASIC, so I can use it to simulate a von Neumann[*] machine

No, that doesn't follow at all. An ASIC doesn't mean a general purpose CPU or FPGA. A chip that only knows how to do, say, video decoding is an example of ASIC. The video chip can't do bitcoin, the bitcoin chip can't do monero. They're not general purpose.

You might be confusing ASICs with FPGAs. You can't reprogram an ASIC, the algorithm is fixed at design time, and the chip built for this single purpose.

> Like, trivially, it's an ASIC, so I can use it to simulate a von Neumann[*] machine

asic does not mean turing complete

good luck simulating a von neumann machine on a sha256 accelerator