Redshift cluster size best practice

Is there a recommendation on the cluster size topology for Redshift?

From previous experience a 2 node cluster is not recommended as you would not get the benefit of distributed computing unless you have a minimum of 3 node.

My question is the following:

Is a 2 node cluster totaling 1TB of memory and similar CPU the equivalent in performance with a 3 node cluster with similar total memory and CPU.

Has anyone done any testing and can share some best practice in terms of cluster topology for redshift?