This paper considers the nonpreemptive scheduling of a given set of jobs on several identical, parallel machines. Each job must be processed on one of the machines. Prior to processing, a job must be loaded (setup) by a single server onto the relevant machine. The paper considers a number of classical scheduling objectives in this environment, under a variety of assumptions about setup and processing times. For each problem considered, the intention is to provide either a polynomial- or pseudo-polynomial-time algorithm, or a proof of binary or unary NP-completeness. The results provide a mapping of the computational complexity of these problems. 2000 Elsevier Science B.V.
