Adapting to Intermittent Faults in Multicore Systems

Future multicore processors will become more susceptible to a variety of hardware failures. In particular, intermittent faults, caused in part by manufacturing process variation or in-progress wear-out, can cause bursts of frequent faults that last from several cycles to several seconds or more. Cost-effective reliability to tolerate intermittent faults will likely require, or be greatly simplified by, the ability to temporarily suspend execution on a core during periods of frequent intermittent faults. We investigate three existing techniques for adapting to the dynamically changing resource availability caused by such core suspension, and demonstrate their different system-level implications. We show that system software reconfiguration has very high overhead for short intermittent faults, that temporarily pausing the execution of a faulty core can lead to cascading livelock, and that using spare cores has high fault-free cost. To remedy these and other drawbacks of current techniques, we propose using a thin hardware/firmware layer to manage an overcommitted system -- one where the OS is configured to use more virtual processors than the number of currently available physical cores. We show that this proposed technique can gracefully degrade performance during intermittent faults of various durations with low overhead, without involving system software, and without requiring spare cores