We present a scalable parallel local search algorithm based on data pa
rallelism. The concept of distributed neighborhood structures is intro
duced, and applied to the Traveling Salesman Problem (TSP). Our parall
el local search algorithm finds the same quality solutions as the clas
sical 2-opt algorithm and has a good speed-up. The algorithm is implem
ented on a Parsytec GCel, consisting of 512 transputers. Its performan
ce is empirically analyzed for TSP instances with several thousands of
cities.