TY - JOUR AB - <jats:p>This paper proposes novel customisable streaming architectures for the NP-hard set covering problem. Our approach covers both exhaustive and genetic algorithms, supporting coarse-grain parallelism and deep pipelines while allowing trade-offs between performance and resource usage. Experiments targeting Maxeler systems show that our FPGA-based designs are more effective than the corresponding multicore software versions. The speed up of the exhaustive algorithm exceeds 250 times, and that of the genetic algorithm exceeds 60 times. Meanwhile, our implementations are more flexible than other FPGA solutions, allowing users to customise parameters at run time without recompilation.</jats:p> AU - Guo,L AU - Thomas,DB AU - Luk,W DO - 10.1145/2641361.2641378 EP - 106 PY - 2013/// SN - 0163-5964 SP - 101 TI - Customisable architectures for the set covering problem T2 - ACM SIGARCH Computer Architecture News UR - http://dx.doi.org/10.1145/2641361.2641378 VL - 41 ER -