We consider a TM application running with N concurrent threads. The TM system is equipped with a thread scheduler that uses some scheduling strategy to adaptively change the number of threads k, with k ≤ N, that are kept running along the application execution.