Question about improving syncing performance for op chains

[yutianwu]

Hi, we are testing the syncing performance for reth on opbnb.

For the opbnb mainnet block range [11035129, 11398984], we tested the performance of reth and op-geth on two machines with identical specifications, each with 32 GB of memory:

• reth: 15.5 hours
• op-geth: 18.4 hours

Reth shows about a 20% improvement compared to op-geth.

In op-geth, we have cached the stateDB after building the payload post forceChoiceUpdate, so we don't need to process the block for newPayload, saving execution time for the same block. You can refer to the implementation here.

The questions are: Do we have a plan to implement a similar caching mechanism in reth, or is it possible to do so based on the current implementation of reth?

Thanks for the reply.