solana requires transactions list all the state/accounts used in the tx metadata (static parallelism), which is super useful for simulation its so easy to build a solana backtesting pipeline that stores all relevant state and efficiency simulate against that state later chains like aptos (ive done a lot of aptos dev) or ethereum dynamically load accounts at runtime which provides nice dev-ex but requires a much more complicated simulation pipeline which potentially needs to save the entire blockchain state at every step to get accurate simulation results playing around with litesvm completely changed my perspective on this

@bl8_runner performance is good enough for me 1m simulated datapoints to generate this chart reran the script and got 577 sim/sec per process, parallelized across all 5 processes per pAMMs ~2,900/s could parallelize further

@bl8_runner **5m datapoints, 1m per pAMM