Making geospatial joins interactive at scale

How does this compare to optimized PostGIS queries in terms of speed?

I was really hoping to see at least a mention of how this compares to other forms of indexing geospatial data.