45th International Conference on Parallel Processing (ICPP), Pennsylvania, United States Of America, 16 - 19 August 2016, pp.46-51
One attractive solution to long simulation time of LDPC codes is to implement inherently parallel decoding algorithms using multicore platforms. In this paper, we present the first OpenMP parallel implementation of LDPC decoding algorithm on a multicore DSP architecture and report its performance. Parallelized Normalized Min-Sum decoding algorithm is implemented on 8-core Texas Instruments (TI) DSP using OpenMP framework. Performance results are obtained by Unified Instrumentation Architecture (UIA). Our results show that the parallelized decoding on 8-core TI DSP achieves more than 6x speedup compared to single-core version.