Thermal and Energy Management of High-Performance Multicores: Distributed and Self-Calibrating Model-Predictive Controller

Bartolini, Andrea; Cacciari, Matteo; Tilli, Andrea; Benini, Luca

doi:10.1109/tpds.2012.117

Cited by 100 publications

(74 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on that, it applies the Borrowing Strategy, adjusting voltage and frequencies at each scheduling tick given the tasks quality requirements, while keeping the mean applied voltage inside the Long Interval VLI lower than VLT C , and the temperature below TLT C . The Short Term Controller can be coupled with state-of-the-art thermal management techniques to handle thermal emergencies [3,12], given the thermal constraint from LTC. The operations performed by the blocks are discussed in more details in the following subsections.…”

Section: Controller Architecturementioning

confidence: 99%

“…The Voltage Selector selects the minimum VLT C , outputted by the Long Term Controller, to guarantee the reliability of the most degraded core (the one with the lowest RP ). Only one VLT C is needed 3 , as the target platform only has a single voltage island. This work can be easily generalized to multiple VLT C 4 Similarly, the Temperature Selector outputs the reference temperature TLT C .…”

Section: Long Term Controllermentioning

confidence: 99%

See 1 more Smart Citation

Workload and user experience-aware dynamic reliability management in multicore processors

Mercati

Bartolini

Paterna

et al. 2013

Proceedings of the 50th Annual Design Automation Conference

Self Cite

View full text Add to dashboard Cite

Reliability is a major concern for nanoscale CMOS circuits. Degradation phenomena such as Electromigration, Negative Bias Temperature Instability, Time Dependent Dielectric Breakdown worsen with transistor scaling. Dynamic Reliability Management (DRM) techniques reduce reliability loss at runtime by constraining operating points, but they face the challenge of reducing user experience degradation while meeting a lifetime target. In this work we propose a sensor based hierarchical controller for multicore processor DRM, exploiting the major gap between the time scales of workload variations and reliability loss. We improve performance and user experience by locally relaxing reliability-induced operating point constraints, while meeting them over the large time windows relevant for reliability. With respect to the state-of-the-art, our solution guarantees timely execution of 100% of latency-critical applications, and have a 4% performance improvement over the whole lifetime.

show abstract

Section: Controller Architecturementioning

confidence: 99%

Section: Long Term Controllermentioning

confidence: 99%

Workload and user experience-aware dynamic reliability management in multicore processors

Mercati

Bartolini

Paterna

et al. 2013

Proceedings of the 50th Annual Design Automation Conference

Self Cite

View full text Add to dashboard Cite

show abstract

“…Thermal management has become a critical issue in the semiconductor industry due to the increasing power density of microprocessors, laser diodes and power amplifiers [1][2][3]. One of the most pressing challenges in current and next-generation electronics is heat removal of wide band gap semiconductors based on gallium nitride (GaN) and silicon carbide (SiC), for high-power radiofrequency (RF) applications [4].…”

Section: Introductionmentioning

confidence: 99%

Design and Modeling of Membrane-Based Evaporative Cooling Devices for Thermal Management of High Heat Fluxes

Salamon²,

Narayanan

et al. 2016

IEEE Trans. Compon., Packag. Manufact. Technol.

View full text Add to dashboard Cite

We present a high-heat-flux cooling device for advanced thermal management of electronics. The device incorporates nanoporous membranes supported on microchannels to enable thin film evaporation. The underlying concept takes advantage of the capillary pressure generated by small pores in the membrane, and minimizes the viscous loss by reducing the membrane thickness. The heat transfer and fluid flow in the device were modeled to determine the effect of different geometric parameters. With the optimization of various parameters, the device can achieve a heat transfer coefficient in excess of 0.05 kW/cm 2 -K while dissipating a heat flux of 1 kW/cm 2 . When applied to power electronics, such as GaN high electron mobility transistors, this membrane-based evaporative cooling device can lower the near junction temperature by more than 40 K compared to contemporary single-phase microchannel coolers.

show abstract

“…Many proposed models target at the on-line temperature regulation applications, in which compact thermal models can estimate or predict the thermal behaviors of a real systems and they can be built in a dynamic way based on the online thermal sensor reading and online power estimation techniques [20][21][22]. Recently a distributed online thermal model was proposed and validated on a realistic many-core system [23][24][25].…”

Section: Introductionmentioning

confidence: 99%

Compact thermal modeling for packaged microprocessor design with practical power maps

et al. 2014

View full text Add to dashboard Cite

Thermal and Energy Management of High-Performance Multicores: Distributed and Self-Calibrating Model-Predictive Controller

Cited by 100 publications

References 37 publications

Workload and user experience-aware dynamic reliability management in multicore processors

Workload and user experience-aware dynamic reliability management in multicore processors

Design and Modeling of Membrane-Based Evaporative Cooling Devices for Thermal Management of High Heat Fluxes

Compact thermal modeling for packaged microprocessor design with practical power maps

Contact Info

Product

Resources

About