Evaluación comparativa del rendimiento entre procesadores CPU y GPU para el problema del viajante

un estudio comparativo computacional utilizando un algoritmo de fuerza bruta

Autores/as

DOI:

https://doi.org/10.47385/cadunifoa.v21.n56.5758

Palabras clave:

Optimización combinatoria, CUDA, Computación paralela, GPU

Resumen

El Problema del Viajante de Comercio (PVC) es uno de los problemas de optimización combinatoria más estudiados en la ciencia de la computación y pertenece a la clase NP-difícil. Este trabajo presenta un análisis comparativo entre implementaciones secuenciales en CPU e implementaciones paralelas en GPU utilizando CUDA para resolver el PVC mediante el algoritmo de fuerza bruta. El estudio se realizó en Python con las bibliotecas NumPy y CuPy en el entorno de Google Colab. Para un grafo con 10 vértices, totalizando 362.880 permutaciones, la implementación en GPU presentó una aceleración (speedup) de 1,30x en relación con la CPU, reduciendo el tiempo de ejecución de 1,3622 segundos a 1,0450 segundos. Los resultados destacan el potencial de la computación paralela en GPU para problemas de optimización combinatoria, especialmente cuando el volumen computacional justifica la sobrecarga de transferencia de datos entre CPU y GPU.

Descargas

Los datos de descargas todavía no están disponibles.

Biografía del autor/a

Vitor Amadeu Souza, Centro Universitário de Volta Redonda UniFOA

PhD candidate in Defense Engineering (IME), MSc in Physics (CBPF), and specialist in several fields, including: Robotics Eng., Electrical Eng., Electronic and Electromechanical Eng., Telecommunications Eng., Control and Automation Eng., Biomedical Eng., Instrumentation Eng., Industry 4.0 Eng., Data Eng., Computer Eng., Software Eng., Networks and Data Security Eng., DevOps Eng., Mechatronics Eng., Embedded Systems Eng., Mechanical Manufacturing Eng., Reliability Eng., Maintenance Eng., Quality Eng., Materials Eng., Production Eng., Product Eng., Packaging Eng., Transport Eng., Highway Eng., Supply Chain Eng., Knowledge Eng., Business Eng., Project Eng., Renewable Energy Eng., Process Eng., Metallurgical Eng., Chemical Eng., and Environmental Eng.

Also specialized in Software Architecture, Cloud Computing, Machine Learning and AI, Internet of Things, Data Science, Full Stack Development, Applied Statistics, Biostatistics, and Project Management. Holds an MBA in Engineering Economics, Data Analysis, and Web 3.0.

Bachelor’s degree in Computer Engineering and licensed to teach Mathematics, Physics, Chemistry, and Philosophy. Also certified as a Systems Analyst and Technician in Electronics, Electrotechnics, Telecommunications, IT, Logistics, Commerce, Business Administration, and Environmental Management.

Professional experience spans several years in electrical, electronic, automation, embedded systems, firmware, and software projects. Develops hardware and software solutions for industrial, automotive, medical, scientific, and commercial automation sectors.

University professor and administrator at Cerne Tecnologia, a company focused on embedded systems development, educational kits, and technological education in areas such as MCUs, FPGAs, programming languages, project design, and PCB layout.

Author of a vast technical and scientific body of work and member of the Brazilian Computer Society (SBC).

Main topics of expertise include: DFT, FFT, PSD, CAN, MODBUS, LIN, TCP/IP, digital filters, digital systems, power systems, Big Data, graphs, PID, fuzzy logic, FPGA, VHDL, Verilog, PLC, DSC, DSP, ARM, frequency inverters, soft starters, solar energy, IoT, LoRa, Java, PHP, JS, REST, Spring Boot, Spark, CSS, SQL, VB, VC#, perceptron, NAO robot, UML, React, among others.

Complete catalog available at Clube de Autores: http://bit.ly/4gwnt78

Citas

APPLEGATE, D. L. et al. The traveling salesman problem: a computational study. Princeton: Princeton University Press, 2007.

GUTIN, G.; PUNNEN, A. P. The traveling salesman problem and its variations. Boston: Springer, 2006. DOI: https://doi.org/10.1007/b101971

KIRK, D. B.; HWEU, W. W. Programming massively parallel processors: a hands-on approach. 3. ed. Burlington: Morgan Kaufmann, 2016.

LAWLER, E. L. et al. The traveling salesman problem: a guided tour of combinatorial optimization. Chichester: Wiley, 1985. DOI: https://doi.org/10.2307/2582681

MITTAL, S.; VETTER, J. S. A survey of CPU-GPU heterogeneous computing techniques. ACM Computing Surveys, v. 47, n. 4, p. 1-35, 2015. Available at: https://dl.acm.org/doi/10.1145/2788396. Accessed on: 20 maio 2025. DOI: https://doi.org/10.1145/2788396

SANDERS, J.; KANDROT, E. CUDA by example: an introduction to general-purpose GPU programming. Boston: Addison-Wesley, 2010.

Binjubeir, Mohammed Mahfoudh & Ismail, Mohd Arfian & Tusher, Ekramul Haque & Aljanabi, Mohammad. (2024). A GPU Accelerated Parallel Genetic Algorithm for the Traveling Salesman Problem. Journal of Soft Computing and Data Mining. 5. 137-150. Available at: http://umpir.ump.edu.my/id/eprint/43901/. Accessed on: 26 maio 2025.

EKER, Musa; ŞEN, Baha; KUMRU, Pinar Yildiz. An efficient solving of the traveling salesman problem: the ant colony system having parameters optimized by the Taguchi method. Turkish Journal of Electrical Engineering and Computer Sciences, v. 21, n. 7, art. 15, 2013. DOI: https://doi.org/10.3906/elk-1109-44. Available at: https://journals.tubitak.gov.tr/elektrik/vol21/iss7/15. Accessed on: 6 jun. 2025.

K. Rocki and R. Suda. High Performance GPU Accelerated Local Optimization in TSP. 2013 IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum, Cambridge, MA, USA, 2013, pp. 1788-1796, doi: 10.1109/IPDPSW.2013.227. Accessed on: 6 mai. 2025. DOI: https://doi.org/10.1109/IPDPSW.2013.227

FUJIMOTO, N. TSUTSUI, S. A highly-parallel TSP solver for a GPU computing platform. In: INTERNATIONAL CONFERENCE ON PARALLEL PROCESSING AND APPLIED MATHEMATICS, 2010. Available at: https://link.springer.com/chapter/10.1007/978-3-642-18466-6_31. Accessed on: 1 mai. 2025.

FOSIN, J.; DAVIDOVIĆ, D.; CARIĆ, T. A GPU implementation of local search operators for symmetric travelling salesman problem. Promet - Traffic & Transportation, v. 25, n. 3, p. 225-234, 2013. Available at: https://hrcak.srce.hr/clanak/164448. Accessed on: 10 mai. 2025. DOI: https://doi.org/10.7307/ptt.v25i3.300

Descargas

Publicado

2026-03-06

Cómo citar

SOUZA, Vitor Amadeu. Evaluación comparativa del rendimiento entre procesadores CPU y GPU para el problema del viajante: un estudio comparativo computacional utilizando un algoritmo de fuerza bruta. Cadernos UniFOA, Volta Redonda, v. 21, n. 56, p. 1–10, 2026. DOI: 10.47385/cadunifoa.v21.n56.5758. Disponível em: https://revistas.unifoa.edu.br/cadernos/article/view/5758. Acesso em: 7 mar. 2026.

Número

Sección

Tecnologia e Engenharias