Cached checksumAddress

[kyscott18]

checksumAddress is a relatively expensive operation because it has to perform the keccak256 hash function on all input.

One option to speed it up would be to add an LRU cache that can potentially skip the expensive hash function when it gets a "hit". This implementation led to a 77x performance improvement.

checksumAddress is also internally used in a few other places, namely decodeAbiParameters. Would you be open to a contribution to add this to viem?

[jxom]

checksumAddress already has an LRU cache.

https://github.com/wevm/viem/blob/main/src/utils/address/getAddress.ts#L34

How does this improve performance by 77x?

Update: I just realised that our LRU implementation doesn’t keep track of the keys touched to flush. Can update to fix.

[kyscott18]

Ah I actually didn't know that you had implemented this already, always one step ahead lol. I was benchmarking against v1.16 which didn't have the LRU cache. Thanks!