Also looking forward to try nostrdb, I’m doing (slowly doing) a Windows / Cmake port. It seems the latest benchmarks are very good @jb55 In my case I have no real DB, and the bottleneck is the time to query relays. For 1 relay only it takes about 4 seconds to get a feed of about 20 follows. But I guess there is not much room here for improvement because of the decentralized nature. So, the idea of having all locally cached in DB, and query asynchronously in background, for example https://github.com/pedro-vicente/nostrdb