Modern real-time systems started leveraging runtime reconfigurable architectures to reduce size, weight, and power, while ensuring predictability. However, the time to reconfigure a platform (referred to as reconfiguration delay) is non-negligible and subject to fluctuations caused by conflicts in accessing shared components during the reconfiguration process. Differently from existing solutions that impact either system resources or system performance, in this letter, we propose an approach, together with its supporting tools, allowing to characterize the reconfiguration delay and to be able to capture and produce evidence of the actual impact of the resource contention. In addition, we report the experiments of the proposed approach in a representative configuration on a Xilinx Zynq Ultrascale+ MPSoC; results show that conflicts can impact the reconfiguration delay at most by 230%.
Analysis of Reconfiguration Delay in Heterogeneous Systems-on-Chip via Traffic Injection
Muttillo V.;
2023-01-01
Abstract
Modern real-time systems started leveraging runtime reconfigurable architectures to reduce size, weight, and power, while ensuring predictability. However, the time to reconfigure a platform (referred to as reconfiguration delay) is non-negligible and subject to fluctuations caused by conflicts in accessing shared components during the reconfiguration process. Differently from existing solutions that impact either system resources or system performance, in this letter, we propose an approach, together with its supporting tools, allowing to characterize the reconfiguration delay and to be able to capture and produce evidence of the actual impact of the resource contention. In addition, we report the experiments of the proposed approach in a representative configuration on a Xilinx Zynq Ultrascale+ MPSoC; results show that conflicts can impact the reconfiguration delay at most by 230%.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
