Re-engineering the ant colony optimization for CMP architectures

Abstract

The Ant Colony Optimization (ACO) is inspired by the behavior of real ants and, as a bioinspired method; its underlying computation is massively parallel by definition. This paper shows re-engineering strategies to migrate the ACO algorithm applied to the Traveling Salesman Problem (TSP) to modern Intel-based multi-and-many-core architectures in a step-bystep methodology. The paper provides detailed guidelines on how to optimize the algorithm for the intra-node (thread and vector) parallelization, showing the performance scalability along with the number of cores on different Intel architectures, reporting up to 5.5x speed-up factor between the Intel Xeon Phi Knights Landing (KNL) and Intel Xeon v2. Moreover, parallel efficiency is provided for all targeted architectures, finding that core load imbalance, memory bandwidth limitations, and NUMA effects on data placement are some of the key factors limiting performance. Finally, a distributed implementation is also presented, reaching up to 2.96x speed-up factor when running the code on 3 nodes over the single-node counterpart version. In the latter case, the parallel efficiency is affected by the synchronization frequency, which also affects the quality of the solution found by the distributed implementation.

Exploiting Heterogeneous Parallelism on Hybrid Metaheuristics for Vector Autoregression Models

“…Metaheuristics can be parallelized in a number of ways [27,28], and there are works on parallelization of each of the basic metaheuristics considered (Local Search [29], Tabu Search [30], Genetic Algorithm [31], Ant Colony [32], ...), and for different types of computational systems (e.g., CMP architectures [33], distributed platforms [34] and GPUs [35,36]). On the other hand, the unified, parameterized schema enables the simultaneous implementation of parallel versions of different basic metaheuristics and their hybridations for different types of computational systems (e.g., shared-memory [11], heterogeneous clusters [12] and many-core systems like GPUs [37]).…”