Manycore Parallel Simulations of Fishing Nets

This article presents the implementation and evaluation of fishing net simulations applying parallel computing on a manycore computing infrastructure. Computational simulations of fishing nets allow researchers to design optimized structures, evaluate materials and different net configurations. Results are important to advance in sustainable fishing practices, reducing the ecological impacts, and ensuring the long-term viability of fisheries resource. A parallel implementation applying the TBB library is developed, following an object-oriented approach to execute in Xeon Phi multicore architectures. The computational efficiency of the developed implementation is studied for different net sizes, considering several metrics for performance evaluation. Results demonstrate a correct scalability of the developed implementation and a proper use of the parallel computing resources. A maximum speedup of 161.81 was achieved when using four threads for each core for simulating a large fishing net.