Update revm

[rakita]

Sorry for the intrusion but this is a interesting benchmark that you have made, it primarily tests and compares initialization part of sputnik and revm.

I didn't spend a lot of time on looking at that part in the start, mostly did optimization on the execution (interpreter) loop, but after tinkering some things on revm/forks (bluealloy/revm#191) branch and adding one more optimization in revm/precomp (bluealloy/revm@d851763) I am getting a lot better times:

execute_contract_method_success_from_rust_evm (1.0s) ...       9_363.419 ns/iter (0.986 R²)
execute_contract_method_reverted_from_rust_evm (1.0s) ...       7_522.533 ns/iter (1.000 R²)
execute_contract_method_success_from_revm (1.0s) ...       2_408.113 ns/iter (1.000 R²)
execute_contract_method_reverted_from_revm (1.0s) ...       2_156.431 ns/iter (1.000 R²)

It would be interesting to see how does it compare on executing heavy contracts, I would expect around x2 speedup but i didn't do the testing :) this is good candidate https://github.com/karalabe/snailtracer

[rakita]

Oh, on snailtracer it is like x5 speedup:

execute_contract_method_success_from_rust_evm (1.8s) ... 261_880_154.000 ns/iter (0.984 R²)
execute_contract_method_success_from_revm (1.1s) ...  49_709_496.686 ns/iter (1.000 R²)

branch: https://github.com/rakita/rust-evm-bench/tree/snailtracer

[cassc]

Thanks @rakita! The performance improvement is very impressive!