The main point was not to fully benchmark and compare both, but just to get a rough sense of whether a Postgres cache was fast enough to be useful in practice. The comparison with Redis was more a crutch to get a sense of that, than really something that pretends to be "rock-solid benchmarking"