Improving Lightning Network Pathfinding Latency by Path Splicing and Other Real-Time Strategy Game Techniques

Author: Rusty Russell 2019-08-08 05:56:13
Published on: 2019-08-08T05:56:13+00:00

• Combined Summary of all posts in thread
• Click here to read original discussion on the lightning-dev mailing list

Summary:

A user on Lightning-dev asked for a reproducible benchmark or further qualification of a statement claiming that commodity hardware would take two seconds to find a route. However, there was no such benchmark available. The user ZmnSCPxj then provided information on an unoptimized c-lightning implementation which returned random routes on the mainnet. On a digital ocean 2-cpu 4GB ram Intel(R) Xeon(R) CPU E5-2630L 0 @ 2.00GHz, it took anywhere between 3 to 347 milliseconds with a mean of 220 milliseconds. On a Raspberry Pi 2B compiled with -O3, it took anywhere between 21 to 3330 milliseconds with a mean of 388 milliseconds.

Updated on: 2023-06-02T19:42:40.155607+00:00