Liked this post. Would love more detail on GPU query offloading and on how different the engine looks compared to one that would run queries on the CPU, as that seems to be the primary innovation
In many ways, this paper is easier to read instead: http://www.csee.usf.edu/~tuy/pub/SSDBM17.pdf
But I'm pretty sure the "moderngpu" version of merge-join is more optimal.