I am trying to figure out how many Bitcoin transactions could be verified per second. Am I correct that signature verification is more expensive than computing Merkletrees and hashes and therefore constitutes the bottleneck of the verification process?
Having read this paper it seems to me that tens of thousands of signatures per second could be verified. Have I misunderstood the results?
As I am only interested in signature verification, let parameters like internet bandwidth and storage requirements to handle the stream of incoming data be outside of the scope of this question.