|
|
| Proposta de dissertação do MEI |
| Título: |
Using Machine Learning to Improve the Scheduling of Computations in Multi-CPU/Multi-GPU Environments |
| Proponente(s): |
Hervé Paulino |
| Créditos: |
42 ECTS |
| Área científica: |
Computer Systems and Networks |
| Início preferencial: |
Qualquer semestre |
| URL: |
|
| Já estão em curso trabalhos preliminares executados pelo alunos: |
|
| Breve descrição: |
This thesis addresses the specification and implementation of a model able to predict the best hardware configuration for executing a computation in a multi-cpu/multi-gpu environment. The work will grow from the existing Marrow platform, which is already able to run computations on those environments.
The work will use Machine Learning techniques to dynamically choose the best hardware configuration for running a given computation with a given input. |
| Observações: |
Papers on Marrow can be found at http://asc.di.fct.unl.pt/~herve/papers/Marrow-Euro-Par2013.pdf
http://asc.di.fct.unl.pt/~herve/papers/Marrow-SAC2014.pdf
https://www.researchgate.net/publication/281313220_Execution_of_Compound_Multi-Kernel_OpenCL_C |
|
|