what’s the advantage of using circuit languages like noir when good zkVMs seem to be on the horizon ie SP1. seems like perf is becoming less of a reason

what’s the advantage of using circuit languages like noir when good zkVMs seem to be on the horizon ie SP1. seems like perf is becoming less of a reason 

https://blog.succinct.xyz/introducing-sp1/

https://material.work


prev: https://persistence.wtf / https://nouns.build

Memory and speed (as it is today)

zkVMs still require a lot of compute and memory where in a mobile app or browser it’s not going idea for an end user product.

Working on building zkTLS infra/products. A ruthless empiricist and meme connoisseur.

/opacity
/eulerlagrange

are there any limitations around using private inputs with zkVMs? seems like most of them are run in a hosted instance and like you mentioned are not feasible to run on a consumer device. guessing this will change with the perf improvements @br1an mentioned

No, not really - everything you need to create a zk RISC Zero proof is completely open. You can make the proof, attest to what you want, and then throw that proof onchain. Zero leakage

Or..... you can use TEEs - this contextual use of a TEE is a lot safer as it's purely for transient privacy