Argumentation Computation with Large Language Models : A Benchmark Study