A fair amount of current High Performance Computing systems take advantage of coprocessors. Most of them use either GPU or FPGA but rarely benefit from both, since the management of such flexible systems is exceedingly challenging. The Cuteforce Analyzer is a multi-purpose cluster system. Different node types allow a variety of cluster configurations to cope with various kinds of tasks. Compute nodes utilize low priced off-the-shelf GPU and/or FPGA as specialized coprocessors to accelerate the execution of algorithms.This paper presents the experiences in implementing the Cuteforce Analyzer and the usage of both coprocessor types in a single cluster system based on MS Windows HPC.