Citus vs regular postgres performance

I was testing citus postgres vs postgres 11 and I tried comparing single postgres 11 t2.medium instance on amazon with:

• t2.medium instance with installed citus postgres 11 (2 working nodes and 1 coordinator on 1 instance)
• 2 worker nodes and 1 coordinator on 3 t2.small instances
• AWS Cloud Formation with citus 8 (3 worker nodes and 1 coordinator all t2.medium instances)

I tried to follow example from demo video https://www.youtube.com/watch?v=g3H4nGsJsl0 and to see results, only with smaller data set (getting data with wget http://examples.citusdata.com/github_archive/github_events-2015-01-01-{0..5}.csv.gz and importing it in databases).

I tried everything from the video, imported data in single postgres instance and all 3 ways of created citus, and postgres instance was faster for me several times then citus. For setting citus instances I used nice instructions from https://docs.citusdata.com/en/v8.1/ (INSTALL section for Ubuntu and AWS Cloud Formation). Setting went without any issues, only thing is I am not getting results as I see on video.

Did anyone else tried it out and compared results? I am quite sure I did something wrong here

Your data probably fits into memory of a single PG node. All your requests are served from memory. You should start seeing performance differences when

1. data is bigger than the available memory of a single server
2. there is some computation on the query that can be paralleled.