General behavioral thermal modeling and characterization for multi-core microprocessor design

Eguia, Thom Jefferson A.; Tan, Sheldon X.-D.; Shen, Ruijing; Pacheco, Eduardo H.; Tirumala, Murli

doi:10.1109/date.2010.5456979

Cited by 12 publications

(11 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Eguia et al [22], [23] improve the previous approach by using a subspace state-space system identification algorithm. As main advantage, it allows to use a generic input vector instead of an impulse stimulus and to identify directly a MIMO system.…”

Section: Related Workmentioning

confidence: 98%

“…technological and design parameters), it enables an effective physical interpretation of the model states with a reduced model order and can capture some real-life behaviours, which are very hard to be caught by simulations: the model learnt is a direct image of the specific device in the deployed environment. Compared to compact model learning strategies [17]- [20], [22], [23], it reduces the parameter set, avoiding over-parametrization and reducing the occurrences of multiple optimums in parameters identification. This is achieved by embedding in the model parameters optimization routine a set of physically meaningful constraints directly derived from thermal laws as the heat equation and the Fourier law.…”

Section: Contributionsmentioning

confidence: 98%

“…As described in Section 1.1 the state-of-the-art approaches [22], [23] to the thermal model identification do not use any prior knowledge of device physics to regularize the thermal model class. These approaches have the main advantage of using out-of-the-box identification tools such as the System Identification Toolbox included in the Maltlab® suite [50].…”

Section: Black Box System Identificationmentioning

confidence: 99%

See 2 more Smart Citations

An Effective Gray-Box Identification Procedure for Multicore Thermal Modeling

Beneventi

Bartolini

Tilli

et al. 2014

IEEE Trans. Comput.

View full text Add to dashboard Cite

Aggressive thermal management is a critical feature for high-end computing platforms, as worst-case thermal budgeting is becoming unaffordable. Reactive thermal management, which sets temperature thresholds to trigger thermal capping actions, is too "nearsighted," and it may lead to severe performance degradation and thermal overshoots. More aggressive proactive thermal managements minimize performance penalty with smooth optimal control. These techniques require knowledge of thermal models, which have to be accurate and simple to make the controls effective, while keeping their complexity limited. In practice, these models are not provided by manufacturers, and in most cases, they strongly depend on the deployment environment. Hence, procedures to automatically derive thermal models in the field are needed. In this paper, we propose a gray-box procedure to learn a compact and physically consistent model for multicore chips. We leverage the physical consistency of the proposed model to tame the model complexity and to face large quantization noise in measurements. We exploit Output Error structures along with Levenberg-Marquardt and Least Squares optimization algorithms. We tackle the problem in a real-life contest: we developed a complete infrastructure for model building and thermal data collection in the Linux environment, and we tested it on an Intel Nehalem-based server CPU.

show abstract

Section: Related Workmentioning

confidence: 98%

Section: Contributionsmentioning

confidence: 98%

Section: Black Box System Identificationmentioning

confidence: 99%

See 1 more Smart Citation

An Effective Gray-Box Identification Procedure for Multicore Thermal Modeling

Beneventi

Bartolini

Tilli

et al. 2014

IEEE Trans. Comput.

View full text Add to dashboard Cite

show abstract

“…Such an option is well suited to deal with process variability and heat sink imperfections. Similarly, since having different fan speeds changes the thermal conductivity of the heat sink, a method like the one in [10] can also help to adapt the thermal model at runtime, or to consider multiple thermal models from which to choose (one for each fan speed).…”

Section: Thermal Modelmentioning

confidence: 99%

“…A more practical alternative is to use some thermal modeling method that relies on real measurements on a specific chip and cooling solution, e.g., [10]. Such an option is well suited to deal with process variability and heat sink imperfections.…”

Section: Thermal Modelmentioning

confidence: 99%

Thermal Safe Power (TSP): Efficient Power Budgeting for Heterogeneous Manycore Systems in Dark Silicon

Pagani

Khdr

Chen

et al. 2017

IEEE Trans. Comput.

View full text Add to dashboard Cite

Chip manufacturers provide the Thermal Design Power (TDP) for a specific chip. The cooling solution is designed to dissipate this power level. But because TDP is not necessarily the maximum power that can be applied, chips are operated with Dynamic Thermal Management (DTM) techniques. To avoid excessive triggers of DTM, usually, system designers also use TDP as power constraint. However, using a single and constant value as power constraint, e.g., TDP, can result in significant performance losses in homogeneous and heterogeneous manycore systems. Having better power budgeting techniques is a major step towards dealing with the dark silicon problem. This paper presents a new power budget concept, called Thermal Safe Power (TSP), which is an abstraction that provides safe power and power density constraints as a function of the number of simultaneously active cores. Executing cores at any power consumption below TSP ensures that DTM is not triggered. TSP can be computed offline for the worst cases, or online for a particular mapping of cores. TSP can also serve as a fundamental tool for guiding task partitioning and core mapping decisions, specially when core heterogeneity or timing guarantees are involved. Moreover, TSP results in dark silicon estimations which are less pessimistic than estimations using constant power budgets.

show abstract