Full text
| |
| Author(s): |
Felipe Ferreira
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | São Carlos. |
| Institution: | Universidade de São Paulo (USP). Instituto de Física de São Carlos (IFSC/BT) |
| Defense date: | 2020-08-25 |
| Examining board members: |
Gonzalo Travieso;
Odemir Martinez Bruno;
Júlio Cezar Estrella;
Edson dos Santos Moreira;
Luis Carlos Trevelin
|
| Advisor: | Gonzalo Travieso |
| Abstract | |
With the conceptual change on processor development that allowed the construction of multi-core processors came the necessity of creating new execution models able to extract all of their processing capacity. Models that use the data flow paradigm are successful in this task, as they allow the available parallelism to be exploited efficiently. The research presented in this monograph aims at the development of a new execution model using the data flow paradigm to efficiently extract the full capacity of today\'s multi-core processors and even of the many-core processors expected to be released soon. A prototype of the model was implemented and analyzed through a comparison with state of the art parallel models using a set of algorithms executed on real and simulated machines. The former allows for an analysis in present-day conditions, while the second allows for some prediction regarding a many-core system environment. The overall result is positive and the developed model is shown to be competitive and even better than some of the state of the art ones, mainly on the many-core region, which is our research goal. (AU) | |
| FAPESP's process: | 15/02014-4 - A data-driven execution model in multi-core processors |
| Grantee: | Felipe Ferreira |
| Support Opportunities: | Scholarships in Brazil - Doctorate (Direct) |