Aging-Aware Parallel Execution

Computation has been pushed to the edge decrease latency and alleviate computational burden of IoT applications in cloud. However, increasing processing demands make necessary employment platforms that exploit thread-level parallelism (TLP). Yet, power heat dissipation rise as TLP inadvertently increases or when is not cleverly exploited, which may be result nonideal use a given parallel program interface (PPI). Besides common issues, such need for more robust sources better cooling, also adversely affects aging, accelerating phenomenons, negative bias temperature instability (NBTI) hot-carrier injection (HCI), further reduces processor lifetime. Hence, considering lifespan an device key, so number times application set execute until its end-of-life maximized, we propose BALDER. It learning framework capable automatically choosing optimal configuration executions (PPI threads) according at hand, aiming maximize tradeoff between aging performance. When executing ten well-known on two multicore embedded architectures, show BALDER can find nearly all our experiments.