FHE computations are hundreds of thousands of times slower than non-encrypted computations, and are non-parallelizable, and Moore's law is long since dead for single-core performance.
I didn't know that. I'm interested in understanding why they aren't parallelizable? I'm surprised that is the case. It's a good point you make, and makes me wonder whether a combination of zk proofs where the inputs to the zk proof are the encrypted inputs of the FHE conjuration. If so, that could be used against a verifier to move the computation off chain. Seems like a complicated circuit you would need to make but if you could do it, you wouldn't need to worry about the computational cost on chain. Thanks for that information, very thought provoking.
