Research Coordinator: Daniel Rossier
High performance embedded systems are based on heterogeneous hardware: processing units (32/64- bit processors/ microcontrollers, DSPs, FPGAs, GPUs, etc.) and peripherals (multimedia, communication, sensors, motor control, etc.).
Reaching a high level of parallelism involves deadling efficiently with software heterogeneity. Our approach consists in partitioning the application on several processing units and using virtualization to run several operating systems on the same hardware.
- Software development for heterogeneous hardware
- Software partitioning on heterogeneous target platforms
- Development of processor clusters and calculation units (DSP, FPGA, etc.)
- Optimal use of hardware resources according to application needs