Abstract
Relocation of partial bitstreams is in the focus of researchers for many years. Several design flows for module relocations have been proposed in the past. In general, these design flows start with a manual identification of equally sized and structured partially reconfigurable regions. Due to the increasing heterogeneity and complexity of modern FPGAs such manual approaches become impractical. This work presents an automated approach for identifying suitable regions for relocatable modules. The algorithm identifies the optimal resource pattern that fits most often on the FPGA device for a given resource requirement. Compared to standard methods trying to find again a manually specified fixed resource pattern, the usage of resource requirements as a starting point augments the amount of identically structured regions to be found.
Original language | English |
---|---|
Title of host publication | 2014 International Conference on ReConFigurable Computing and FPGAs (ReConFig14) |
Number of pages | 6 |
Publisher | IEEE |
Publication date | 01.12.2014 |
Pages | 1-6 |
Article number | 7032533 |
ISBN (Print) | 978-1-4799-5943-3 |
ISBN (Electronic) | 978-1-4799-5944-0 |
DOIs | |
Publication status | Published - 01.12.2014 |
Event | 2014 International Conference on Reconfigurable Computing and FPGAs - Cancun, Mexico Duration: 08.12.2014 → 10.12.2014 Conference number: 110856 |
Research Areas and Centers
- Research Area: Intelligent Systems
- Centers: Center for Artificial Intelligence Luebeck (ZKIL)
DFG Research Classification Scheme
- 409-04 Operating, Communication, Database and Distributed Systems