Our circuit is already pretty optimized unfortunately.
60% of cycles are spent calculating the merkle root, 30% on deserializing the data into risc0.
Maybe we can use the older risc0 custom sterilization instead of serde.
I can align the data a bit better, maybe I can optimize calc_subtree_hash a bit, but not much.
Our circuit is already pretty optimized unfortunately.
60% of cycles are spent calculating the merkle root, 30% on deserializing the data into risc0. Maybe we can use the older risc0 custom sterilization instead of serde. I can align the data a bit better, maybe I can optimize
calc_subtree_hash
a bit, but not much.