Stochastic modeling of a thermally-managed multi-core system

Jung, Hwisung; Peng, Rong; Pedram, Massoud

doi:10.1145/1391469.1391657

Cited by 38 publications

(13 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One possible approach to engage more than one response for thermal management is to determine a crossover point between the thermal management techniques as in [15] where fetch gating and dynamic voltage scaling are combined. Similarly Jung et al [8] employ exhaustive simulation and stochastic modeling to determine the best power management policy with two configuration parameters, namely, cache size and frequency. It is unclear how to determine the cross-over point when multiple mechanisms are employed.…”

Section: Related Workmentioning

confidence: 99%

“…It is unclear how to determine the cross-over point when multiple mechanisms are employed. Similarly, the large configuration space for multiple mechanisms makes exhaustive simulation like [8] infeasible. In our approach, we view the thermal management problem as a configuration space exploration problem and design an efficient online technique to determine the configuration that results in maximum performance for the workload under a given temperature constraint.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Dynamic thermal management via architectural adaptation

Jayaseelan

Mitra

2009

Proceedings of the 46th Annual Design Automation Conference

View full text Add to dashboard Cite

Exponentially rising cooling/packaging costs due to high power density call for architectural and software-level thermal management. Dynamic thermal management (DTM) techniques continuously monitor the on-chip processor temperature. Appropriate mechanisms (e.g., dynamic voltage or frequency scaling (DVFS), clock gating, fetch gating, etc.) are engaged to lower the temperature if it exceeds a threshold. However, all these mechanisms incur significant performance penalty. We argue that runtime adaptation of micro-architectural parameters, such as instruction window size and issue width, is a more effective mechanism for DTM. If the architectural parameters can be tailored to track the available instruction-level parallelism of the program, the temperature is reduced with minimal performance degradation. Moreover, synergistically combining architectural adaptation with DVFS and fetch gating can achieve the best performance under thermal constraints. The key difficulty in using multiple mechanisms is to select the optimal configuration at runtime for time varying workloads. We present a novel software-level thermal management framework that searches through the configuration space at regular intervals to find the best performing design point that is thermally safe. The central components of our framework are (1) a neural-network based classifier that filters the thermally unsafe configurations, (2) a fast performance prediction model for any configuration, and (3) an efficient configuration space search algorithm. Experimental results indicate that our adaptive scheme achieves 59% reduction in performance overhead compared to DVFS and 39% reduction in overhead compared to DVFS combined with fetch gating.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Dynamic thermal management via architectural adaptation

Jayaseelan

Mitra

2009

Proceedings of the 46th Annual Design Automation Conference

View full text Add to dashboard Cite

show abstract

“…Some fast system level temperature analysis techniques are proposed, e.g., [7] and [8] which are efficient to be used inside the temperature aware system level optimization loop. Several temperature aware system level design approaches, e.g., [9] and [10], are proposed in which decisions are taken on-line, based on the actual chip temperature information. For online temperature monitoring, sensors have been used together with techniques for collecting and analyzing their values with adequate accuracy, e.g., [11].…”

Section: Introductionmentioning

confidence: 99%

On-line Temperature-Aware Idle Time Distribution for Leakage Energy Optimization

Bao

Alexandru

Eles

et al. 2011

2011 Sixth IEEE International Symposium on Electronic Design, Test and Application

View full text Add to dashboard Cite

With new technologies, temperature has become an important issue to be considered at system level design. In this paper, we address the issue of leakage energy optimization through temperature aware idle time distribution (ITD). We propose an on-line ITD technique for leakage energy consumption minimization, where both static and dynamic idle time are considered. Experimental results have demonstrated that an important amount of leakage energy reduction can be achieved by applying our ITD techniques.

show abstract

“…Jung [12] proposed a temperature management approach based on a Markovian decision process aiming at minimizing energy under temperature constraints. The abstract model of the managed system is constructed at design time, based on a very simple temperature model.…”

Section: Introductionmentioning

confidence: 99%

On-line thermal aware dynamic voltage scaling for energy optimization with frequency/temperature dependency consideration

Bao

Alexandru

Eles

et al. 2009

Proceedings of the 46th Annual Design Automation Conference

View full text Add to dashboard Cite

With new technologies, temperature has become a major issue to be considered at system level design. Without taking temperature aspects into consideration, no approach to energy or/and performance optimization will be sufficiently accurate and efficient. In this paper we propose an on-line temperature aware dynamic voltage and frequency scaling (DVFS) technique which is able to exploit both static and dynamic slack. The approach implies an offline temperature aware optimization step and on-line voltage/frequency settings based on temperature sensor readings. Most importantly, the presented approach is aware of the frequency/temperature dependency, by which important additional energy savings are obtained.

show abstract

Stochastic modeling of a thermally-managed multi-core system

Cited by 38 publications

References 13 publications

Dynamic thermal management via architectural adaptation

Dynamic thermal management via architectural adaptation

On-line Temperature-Aware Idle Time Distribution for Leakage Energy Optimization

On-line thermal aware dynamic voltage scaling for energy optimization with frequency/temperature dependency consideration

Contact Info

Product

Resources

About