Data-driven multi-objective optimization of ML inference hardware configurations for energy, performance and cost
Authors:
- Joel CastañoJCJoel Castaño
- Jaime BustilloJBJaime Bustillo
- Xavier FranchXFXavier Franch
- Silverio Martínez-FernándezSMSilverio Martínez-Fernández
Open Access
- 31 March 2026
- journal article
- Published by Elsevier in Sustainable Computing: Informatics and Systems
- Vol. 50
Abstract
No abstract availableFunding Information
- Agencia Estatal de Investigación (PID2024-156019OB-I00)
- Agència de Gestió d'Ajuts Universitaris i de Recerca (2025-PROD-00236)
- Ministerio de Ciencia, Innovación y Universidades
- European Commission
- European Regional Development Fund
This publication has 17 references indexed in Scilit:
- Energy-efficient neural network training through runtime layer freezing, model quantization, and early stoppingComputer Standards & Interfaces, 2024
- A review of green artificial intelligence: Towards a more sustainable futureNeurocomputing, 2024
- SLO-Aware GPU DVFS for Energy-Efficient LLM Inference ServingIEEE Computer Architecture Letters, 2024
- EALI: Energy-aware layer-level scheduling for convolutional neural network inference services on GPUsNeurocomputing, 2022
- The Carbon Footprint of Machine Learning Training Will Plateau, Then ShrinkComputer, 2022
- Green AICommunications of the ACM, 2020
- InferLine2020
- MLPerf: An Industry Standard Benchmark Suite for Machine Learning PerformanceIEEE Micro, 2020
- The Green500 List: Encouraging Sustainable SupercomputingComputer, 2007
- A fast and elitist multiobjective genetic algorithm: NSGA-IIIEEE Transactions on Evolutionary Computation, 2002