Some implementation aspects of Concurrent Clean on distributed memory architectures. John H. G. van Groningen Abstract. We have implemented a code generator and runtime system that can be used to simulate parallel execution of Concurrent Clean programs on a single Macintosh computer. This code generator generates machine code and is an extension of the (sequential) ABC code generator for the MC680x0 and SPARC processors. This implementation is discussed briefly. The largest part of the paper describes two aspects of the implementation in detail. The first one is how to let a process wait until a node is overwritten with its head normal form. Several possible solutions are described. The other one is a description of an efficient graph copying algorithm. This graph copying algorithm can be used to send graphs to remote processors using a more compact representation. Finally some results are presented.