Reliability Assessment of Many-Core Dynamic Thermal Management

Power density can limit the amount of energy a many-core can consume. A many-core running in its maximum performance can lead to safe temperature violation and reliability issues. The literature presents dynamic thermal management (DTM) techniques that guarantee system operation according to safe temperature restrictions. The state-of-art also targets dynamic reliability management (DRM), aiming for longer lifetime reliability, using the same actuation knobs used to control the temperature. This work assesses the reliability of DTM techniques applied in high and dynamic workloads, showing that DTMs keep the temperature within safe limits and increase system lifetime reliability.