Report · 2021
On-Chain Spatial Computation Benchmark
We benchmarked a quadtree-based spatial query engine for EVM chains -- point and range queries over reference GeoJSON datasets, measured locally and on the Celo Alfajores testnet, with comparisons against FOAM's crypto-spatial coordinates and gas-cost measurements for insertion. The benchmark work was carried out for us by Tero Labs.
Purpose
Before pushing spatial computation on-chain, we needed numbers -- what do spatial queries actually cost on an EVM, and how do they scale?
Approach
Benchmarks of a quadtree-based spatial query engine -- point and range queries over reference GeoJSON datasets on a local EVM chain and the Celo Alfajores testnet, compared against FOAM's crypto-spatial coordinates, with gas costs for insertion. We commissioned the benchmark work from Tero Labs, whose contribution we gratefully acknowledge.
Status
Working report, 2021, hosted here as a PDF; includes some unfinished sections.