RT Conference Proceedings
T1 A scheduling theory framework for GPU tasks efficient execution
A1 Lázaro Muñoz, Antonio José
A1 López Albelda, Bernabé
A1 González-Linares, José María
A1 Guil-Mata, Nicolás
K1 Arquitectura de redes informáticas - Congresos
AB Concurrent  execution  of  tasks  in  GPUs can  reduce  the  computation  time  of  a  workload  byoverlapping  data  transfer  and  execution  commands.However it is difficult to implement an efficient run-time scheduler that minimizes the workload makespanas many execution orderings should be evaluated.  Inthis  paper,  we  employ  scheduling  theory  to  build  amodel  that  takes  into  account  the  device  capabili-ties, workload characteristics, constraints and objec-tive  functions.    In  our  model,  GPU  tasks  schedul-ing  is  reformulated  as  a  flow  shop  scheduling  prob-lem, which allow us to apply and compare well knownmethods already developed in the operations researchfield.  In addition we develop a new heuristic, specif-ically  focused  on  executing  GPU  commands,   thatachieves better scheduling results than previous tech-niques.  Finally, a comprehensive evaluation, showingthe  suitability  and  robustness  of  this  new  approach,is conducted in three different NVIDIA architectures(Kepler, Maxwell and Pascal).
YR 2018
FD 2018-07-16
LK https://hdl.handle.net/10630/16279
UL https://hdl.handle.net/10630/16279
LA eng
NO Proyecto TIN2016- 0920R, Universidad de Málaga (Campus de Excelencia Internacional Andalucía Tech) y programa de donación de NVIDIA Corporation.
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 19 ene 2026