Long-term reliability of processors in SoCs and NoCs is experiencing growing attention lately, since decreasing feature sizes and increasing temperatures have a negative influence on the lifespan. Recent work suggests an interplay between power management and reliability, since power management strategies affect the temperature of processors. Power management strategies are examine, which target to actively influence the long-term reliability of a multi-core processor, in integrated systems as e.g. for SoCs and NoCs.

The approach shows that dynamic parallelism can improve the reliability of multi-core systems significantly. First results were achieved by simulating a multi-processor using the Self Distributing Virtual Machine (SDVM) as a basis.