Optimizing Parallel Bitonic Sort
IPPS '97 Proceedings of the 11th International Symposium on Parallel Processing
An improvement of Bitonic sorting for parallel computing
ICCOMP'05 Proceedings of the 9th WSEAS International Conference on Computers
ACM Transactions on Programming Languages and Systems (TOPLAS)
| Hi-index | 0.00 |
We address the problem of sorting a large number N ofkeys on a MasPar MP-1 parallel SIMD machine of moderatesize P where the processing elements (PEs) are interconnectedas a toroidal mesh and have 16KB local storageeach. We present a comparative study of implementationsof the following deterministic oblivious sorting methods:Bitonic Sort, Odd-Even Merge Sort, and FastSort. Wesuccessfully use the guarded split&merge operation introducedby Rüb. The experiments and investigations in a simple,parameterized, analytical model show that, with thisoperation, from a certain ratio N=P upwards both Odd-Even Merge Sort and FastSort become faster on averagethan the up to the present fastest, sophisticated implementationof Bitonic Sort by Prins. Though it is not as efficientas Odd-Even Merge Sort, FastSort is to our knowledge thefirst method specially tailored to the mesh architecture thatcan be, when implemented, competitive on average with amesh-adaptation of Bitonic Sort for large N=P.