Dynamic Thermal Management Research Articles

Accurate On-Chip Thermal Peak Detection Based on Heuristic Algorithms and Embedded Temperature Sensors

The reliability and lifetime of systems-on-chip (SoCs) are being seriously threatened by thermal issues. In modern SoCs, dynamic thermal management (DTM) uses the thermal data captured by thermal sensors to constantly track the hot spots and thermal peak locations in real time. Estimating peak temperatures and the location of these peaks can play a crucial role for DTM systems, as temperature underestimation can cause SoCs to fail and have shortened lifetime. In this paper, a novel sensor allocation algorithm (called thermal gradient tracker, TGT), based on the recursive elimination of regions that likely do not contain any thermal peaks, is proposed for determining regions that potentially contain thermal peaks. Then, based on an empirical source temperature detection technique called GDS (gradient direction sensor), a hybrid algorithm for detecting the position and temperature of thermal peaks is also proposed to increase the accuracy of temperature sensing while trying to keep the number of thermal sensors to a minimum. The essential parameters, H and R, of the GDS technique are determined using an automated search algorithm based on simulated annealing. The proposed algorithm has been applied in a system-on-chip (SoC) in which four heat sources are present, and for temperatures ranging between 45 °C and 115 °C, in a chip area equal to 25 mm2. The simulation results show that our proposed sensor allocation scheme can detect on-chip peaks with a maximum error of 1.48 °C and an average maximum error of 0.49 °C by using 15 thermal sensors.

Hot-Trim: Thermal and Reliability Management for Commercial Multicore Processors Considering Workload Dependent Hot Spots

This work proposes a new dynamic thermal and reliability management framework via task mapping and migration to improve thermal performance and reliability of commercial multi-core processors considering workload-dependent thermal hot spot stress. The new method is motivated by the observation that different workloads activate different spatial power and thermal hot spots within each core of processors. Existing run-time thermal management, which is based on on-chip location-fixed thermal sensor information, can lead to suboptimal management solutions as the temperatures provided by those sensors may not be the true hot spots. The new method, called Hot-Trim, utilizes a machine learning-based approach to characterize the power density hot spots across each core, then a new task mapping/migration scheme is developed based on the hot spot stresses. Compared to existing works, the new approach is the first to optimize VLSI reliabilities by exploring workload-dependent power hot spots. The advantages of the proposed method over the Linux baseline task mapping and the temperature-based mapping method are demonstrated and validated on real commercial chips. Experiments on a real Intel Core i7 quad-core processor executing PARSEC-3.0 and SPLASH-2 benchmarks show that, compared to the existing Linux scheduler, core and hot spot temperature can be lowered by 1.15 1.31∘C. In addition, Hot-Trim can improve the chip’s EM, NBTI and HCI related reliability by 30.2%, 7.0% and 31.1% respectively compared to Linux baseline without any performance degradation. Furthermore, it improves EM and HCI related reliability by 29.6% and 19.6% respectively, and at the same time even further reduces the temperature by half a degree compared to the conventional temperature-based mapping technique.

An Adaptive Multispectral Mechano-Optical System for Multipurpose Applications.

Mechano-optical systems with on-demand adaptability and a broad spectrum from the visible to microwave are critical for complex multiband electromagnetic (EM) applications. Most existing material systems merely have dynamic optical or microwave tunability because their EM wave response is strongly wavelength-dependent. Inspired by cephalopod skin, we develop an adaptive multispectral mechano-optical system based on bilayer acrylic dielectric elastomer (ADE)/silver nanowire (AgNW) films, which reconfigures the surface morphology between wrinkles and cracks via mechanical contraction and stretching. Such morphological evolution regulates the direct transmission/reflection and scattering behavior of visible-infrared light and simultaneously alters the conductive network in a AgNW film to influence its microwave characteristics. The designed system features switching between visible-infrared-microwave transparency and opacity, continuous regulation, wide spectral window (0.38-15.5 μm and 24,200-36,600 μm), excellent recyclability (500 times), and rapid response time (<1 s). These grant the system great potential as platforms for various promising applications such as smart windows, switchable EM devices, dynamic thermal management, adaptive visual stealth, and human motion detection.

Bifunctional silicone triggered long-range crosslinking phenolic aerogels with flexibility and thermal insulation for thermal regulation

Lightweight and flexible aerogels are highly demanded for thermal management in the burgeoning aerospace and artificial intelligence (AI) fields. However, the design and preparation of thermal insulating aerogels integrating good flexibility, low density, and superior thermostability through environmentally friendly methods remains challengeable. Herein, by introducing bifunctional silicone, a strategy of chain elongation assisted with pore structure adjustment is developed to achieve the flexibility of thermostable phenolic resin aerogel (PRA). The silicone introduction effectively suppresses the drying shrinkage, thus enabling the employment of ambient pressure drying (APD). The obtained flexible PRA combining high softness (elastic modulus 0.05 kPa), superior deformability (cyclic compressive strain > 60%, damage strain up to 80%), and good fatigue-resistance (compression at 40% strain > 1000 cycles) exhibits low bulk density (0.058 g·cm−3), and low thermal conductivity (0.033 W·m−1·K−1). With these superior properties, it can effectively serve for static thermal insulation below 200 °C with a thickness of 2–3 mm; more attractively, it can perform dynamic thermal management as a regulator through compressing and releasing (eg. temperature 41.9–57.6 °C, 0–40% compression strain, thickness of 30 mm). These merits make the material promising for AI thermal management in fields like outdoors, aerospace, and infrared camouflage.

Nanoengineering Natural Leather for Dynamic Thermal Management and Electromagnetic Interference Shielding.

Unpredictable and extreme weather conditions, along with increasing electromagnetic pollution, have resulted in a significant threat to human health and productivity, causing irreversible damage to society's well-being and economy. However, existing personal temperature management and electromagnetic protection materials lack adaptability to dynamic environmental changes. To address this, a unique asymmetric bilayer leather/a-MWCNTs/CA fabric isdeveloped by vacuum-infiltrating interconnected a-MWCNTs networks into natural leather's microfiber backbone and spraying porous acetic acid (CA) on the reverse side. Such fabric achieves simultaneous passive radiation cooling, heating, and anti-electromagnetic interference functions without external energy input. The fabric's cooling layer has high solar reflectance (92.0%) and high infrared emissivity (90.2%), providing an average subambient radiation cooling effect of 10 °C, while the heating layer has high solar absorption (98.0%), enabling excellent passive radiative heating and effective compensation for warming via Joule heating. Additionally, the fabric's 3D conductive a-MWCNTs network provides electromagnetic interference shielding effectiveness of 35.0dB mainly through electromagnetic wave absorption. This multimode electromagnetic shielding fabric can switch between cooling and heating modes to adapt to dynamic cooling and heating scenarios, providing a new avenue for sustainable temperature management and electromagnetic protection applications.

Flexible Active Peltier Coolers Based on Interconnected Magnetic Nanowire Networks.

Thermoelectric energy conversion based on flexible materials has great potential for applications in the fields of low-power heat harvesting and solid-state cooling. Here, we show that three-dimensional networks of interconnected ferromagnetic metal nanowires embedded in a polymer film are effective flexible materials as active Peltier coolers. Thermocouples based on Co-Fe nanowires exhibit much higher power factors and thermal conductivities near room temperature than other existing flexible thermoelectric systems, with a power factor for Co-Fe nanowire-based thermocouples of about 4.7 mW/K2m at room temperature. The effective thermal conductance of our device can be strongly and rapidly increased by active Peltier-induced heat flow, especially for small temperature differences. Our investigation represents a significant advance in the fabrication of lightweight flexible thermoelectric devices, and it offers great potential for the dynamic thermal management of hot spots on complex surfaces.

A multi-mode cellulose acetate/MXene Janus film with structure enhanced self-reflection, selective emission and absorption for cooling and heating

Passive radiative thermal management has been recognized as one of the most promising energy-saving technologies to provide thermal comfortable environment. However, static function and complex production process limit its application. Herein, we propose a simple strategy to achieve the integration of radiative cooling, radiative heating, and active electro-thermal heating. It is worth stressing that the radiative cooling is only based on the self-reflective structure and self-emissive molecular property of cellulose acetate electrospun membrane. In cooling mode, the average temperature of Janus film outdoor is 9.4 °C lower than commercial cotton fabric. Meanwhile, radiative heating and active heating stem from the photothermal and electrothermal performance of MXene. In heating mode, the average temperature of Janus film is 40 °C higher than cotton fabric. Flipping the Janus film allows for dynamic thermal management. In short, the scalable and universal strategy paves the way for the development of dynamic thermal-regulating technology.

A Comprehensive Overview of Basic Research on Human Thermal Management in Future Mobility: Considerations, Challenges, and Methods

Thermal management in automobiles is necessary to minimize energy usage while achieving a high level of occupant thermal perception. As the freedom of in-vehicle activity increases with autonomous driving, considering convenience becomes even more important, and, at the same time, the need for thermal management in electric vehicles is expected to increase. While it is necessary to consider the characteristics of the future mobility environment, there is still a lack of research that takes into account these changes in thermal management and proposes future research directions. Therefore, the purpose of this study is to explore basic research directions based on national R&amp;D project cases and to provide a comprehensive overview. The environmental changes that need to be considered in thermal management research include the diversification of future transportation types and usage purpose diversification, the characteristics of electric vehicles, changes in the interior and exterior design of autonomous vehicles, personalized air conditioning environments, and dynamic thermal management according to occupant in-vehicle activity. This study provides an understanding of the overall field, and can help identify challenges, solutions, and ideas. Although this study provides conceptual considerations for research directions, future research is needed to identify detailed factors related to technology, environment, and human factors.

Twist-Induced Near-Field Radiative Heat Switch in Hyperbolic Antiferromagnets

We study the twisted control of the near-field radiative heat transfer between two hyperbolic antiferromagnetic insulators under external magnetic fields. We show that the near-field heat flux can be affected by both the twist angle $\theta$ and the magnitude of the applied magnetic field with different broken symmetries. Irrespective of twist angle, the external magnetic field causes the radiative heat flux to change nonmonotonically, and the minimum heat flux can be found with the magnetic fields of about 1.5 T. Such nonmonotonic behavior is due to the fact that the magnetic field can radically change the nature of the magnon polaritons with time reversal symmetry breaking. The field not only affects the topological structure of surface magnon polaritons, but also induces the volume magnon polaritons that progressively dominate the heat transfer as the field increases. We further propose a twist-induced thermal switch device with inversion symmetry breaking, which can severely regulate radiative heat flux through different magnetic fields. Our findings account for a characteristic modulation of radiative heat transfer with implications for applications in dynamic thermal management.

NFRHT modulation between graphene/SiC core-shell and hBN plate through strain.

We numerically investigate the near-field radiative heat transfer (NFRHT) between a graphene/SiC core-shell (GSCS) nanoparticle and a hexagonal boron nitride (hBN) plate. By applying a compressive strain to the hBN plate, its hyperbolic modes can be tuned. Consequently, the hyperbolic phonon polaritons (HPPs) of hBN and the high-frequency localized surface resonance (LSR) of GSCS nanoparticle can couple and decouple, thus allowing for the active control of NFRHT. Furthermore, we predict that, combining with the effect of the chemical potential of graphene shell on NFRHT, a thermal rectification ratio of up to 13.6 can be achieved. This work enriches the phonon-polariton coupling mechanism and also facilitates dynamic thermal management at the nanoscale.

Recent advances in dynamic dual mode systems for daytime radiative cooling and solar heating.

Thermal management, including heating and cooling, plays an important role in human productive activities and daily life. Nevertheless, in the actual environment, almost all the ambient scenarios come with the challenge that the objects are located in a quite dynamic and variable environment, which includes fluctuations in aspects such as space, time, sunlight, season, and temperature. It is imperative to develop low-energy or even zero-energy thermal-management technologies with renewable and clean energy. In this review, we summarised the latest technological advances and the prospects in this burgeoning field. First, we present the fundamental principles of the daytime passive radiative cooling (PDRC) thermal management device. Next, In the domain of dual-mode systems, they are classified into various types based on the diverse mechanisms of transitioning between cooling and heating states, including electrical responsive, mechanical responsive, temperature responsive, and solution responsive. Furthermore, we conducted an in-depth analysis of the principles and design methodologies associated with these categories, followed by a comparative assessment of their performance in radiative cooling and solar heating applications. Finally, this review presents the challenges and opportunities of dynamic dual mode thermal management, while also identifying future directions.

Dynamic thermal management of flashing spray cooling by the frequency conversion of compressor

Dynamic thermal management of electronic equipment is of practical importance to adapt to the quickly changed heat load at local spot. A closed-loop experimental system was constructed to study the heat transfer mechanism and cooling performance of flashing spray cooling by R410A on a smooth flat copper surface in steady-state and varying operating conditions, especially the effect of frequency conversion of compressor. The results indicated that the superheat of 4 ℃ and 22 ℃ can be regarded as the transition points between three heat transfer stages of steady R410A flashing spray cooling: single-phase heat transfer, nucleate boiling, and transition boiling. To achieve the best cooling performance, superheat should be between 4 ℃ and 22 ℃ to keep heat transfer in stage II. The increase of compressor frequency can improve the cooling performance of spray system effectively. When the compressor frequencies are 10 Hz, 15 Hz and 20 Hz, CHF can reach 162.9, 168.3 and 175.0 W/cm2, and the maximum heat transfer coefficient is around 60.5, 80.6, and 90.4 kW/(m2·K), separately, with surface temperature below 35 ℃, 30 ℃ and 25 ℃. At the same heat flux, the power consumptions for the system to reach steady-state at 15 Hz and 20 Hz are 12.2% lower but 5.9% higher than that at 10 Hz, which provides a feasible energy-saving strategy for spray cooling system in the practical thermal management of electronic equipment.

Deep reinforcement learning towards real-world dynamic thermal management of data centers

Deep Reinforcement Learning has been increasingly researched for Dynamic Thermal Management in Data Centers. However, existing works typically evaluate the performance of algorithms on a specific task, utilizing models or data trajectories without discussing in detail their implementation feasibility and their ability to deal with diverse work scenarios. The lack of these works limits the real-world deployment of Deep Reinforcement Learning. To this end, this paper comprehensively evaluates the strengths and limitations of state-of-the-art algorithms by conducting analytical and numerical studies. The analysis is conducted in four dimensions: algorithms, tasks, system dynamics, and knowledge transfer. As an inherent property, the sensitivity to algorithms settings is first evaluated in a simulated data center model. The ability to deal with various tasks and the sensitivity to reward functions are subsequently studied. The trade-off between constraints and power savings is identified by conducting ablation experiments. Next, the performance under different work scenarios is investigated, including various equipment, workload schedules, locations, and power densities. Finally, the transferability of algorithms across tasks and scenarios is also evaluated. The results show that actor-critic, off-policy, and model-based algorithms outperform others in optimality, robustness, and transferability. They can reduce violations and achieve around 8.84% power savings in some scenarios compared to the default controller. However, deploying these algorithms in real-world systems is challenging since they are sensitive to specific hyperparameters, reward functions, and work scenarios. Constraint violations and sample efficiency are some aspects that are still unsatisfactory. This paper presents our well-structured investigations, new findings, and challenges when deploying deep reinforcement learning in Data Centers.

A Nearest-Neighbor-Based Thermal Sensor Allocation and Temperature Reconstruction Method for 3-D NoC-Based Multicore Systems

To avoid overheating of 3-D network-on-chip (NoC)-based multicore systems, many researchers have used dynamic thermal management (DTM) techniques, which need embedded thermal sensors to provide accurate temperature information. However, only a few sensors can be embedded due to the limited hardware cost. So, it is crucial to find an appropriate way to allocate number-limited sensors at design time and reconstruct full-chip temperature accurately using the limited temperature information. However, the relationship between the non-sensor-allocated nodes and the sensor-allocated nodes modeled by the existing methods has a deviation from the actuality, which leads to an inaccurate temperature reconstruction. Another problem is that the existing methods depend highly on the training data. The estimation error can be significant when the running application’s traffic characteristic differs from the one in the offline phase. This article presents a sensor allocation method based on the cores’ spatial correlation, which is not dependent on the training data. Our allocation method contains two stages: 1) using our nearest-neighbor-based initialization algorithm to allocate sensors preliminarily and 2) using genetic algorithm (GA) to optimize the initial allocation. Besides, we use artificial neural network (ANN) to reconstruct the full-chip temperature. Compared with the state-of-the-art methods, our method can improve the average accuracy of the estimated temperature under different scenarios by 17.60%–88.63%. What is more, our approach has high flexibility and can adapt to different application scenarios with high accuracy with only one offline training.

NeuroMap: Efficient Task Mapping of Deep Neural Networks for Dynamic Thermal Management in High-Bandwidth Memory

High-bandwidth memory (HBM) offers breakthrough memory bandwidth through its vertically stacked memory architecture and through-silicon via (TSV)-based fast interconnect. However, the stacked architecture leads to high-power density causing thermal issues when running modern memory-hungry workloads such as deep neural networks (DNNs). Prior works on dynamic thermal management (DTM) of 3-D DRAM do not consider the physical structure of HBM and often lead to heavy DTM-induced performance penalty. We propose an application-aware efficient task mapping and migration-based DTM policy that maps DNN instances to cores through exploiting the channel layout of HBM and leveraging the significant temperature gradient across DRAM dies while making thermal decisions. We utilize the variation in the memory access behavior of DNN layers and attempt to minimize stalling due to thermal hotspots in the HBM stack. We also use application-aware dynamic voltage and frequency scaling (DVFS) and DRAM low-power states to further improve performance. Experimental results on workloads comprising seven popular DNNs show that NeuroMap results in an average execution time and memory energy reduction of 39% and 40%, respectively, over state-of-the-art DTM mechanisms.

DRL-S: Toward safe real-world learning of dynamic thermal management in data center

Deep Reinforcement Learning has been researched for Dynamic Thermal Management in Data Centers. An objective of Dynamic Thermal Management is to minimize power consumption while satisfying a set of hard and soft constraints. However, there is a lack of research on the safety issues of applying Deep Reinforcement Learning in Data Centers during real-world learning, limiting its deployment. To this end, this paper proposes a new method named DRL-S, which can solve both hard and soft constraints simultaneously during the learning stage. In addition to optimizing the main objective, Lagrangian-based Constrained Deep Reinforcement Learning and Reward Shaping enforce policies to satisfy soft constraints through extensive online learning. However, the policy typically fails to satisfy the hard constraints due to the random sampling of actions to encourage exploration and the imperfect policy at the initial online learning stage. To ensure hard constraints are satisfied, we further propose to utilize parameterized Shielding, integrating the approximation of the system dynamics and the projection of the action space to predict the safety of candidate actions and provide backup actions when necessary. Results show that the Lagrangian-based method and Reward Shaping can gradually learn policies to reduce soft constraint violations. The former can better balance the relationship between the main objective and violations by updating Lagrangian multipliers. DRL-S can also effectively avoid extreme temperatures without affecting the normal learning process of vanilla algorithms. The asymptotic power consumption is more than 12% lower than the baseline controller.

(Invited, Digital Presentation) ML-Based Fast on-Chip Transient Thermal Simulation for Heterogeneous 2.5D/3D IC Designs

Large 3DIC designs with multiple chips requireseveral iterations of transient thermal analysis particularly forfine-grain on-chip dynamic thermal management. This requiresa fast thermal analysis technology as opposed to traditionalCFD/FEA based methods which have severe runtime/capacitylimitations for large chips (e.g., 2cmx2cm) in 3DIC whilegenerating fine grained (e.g., 10umx10um) transient thermalresponse. The fast transient thermal analysis is based on the ideaof combining the global, intermediate, and local transientresponse curves generated from an ML-predictor. The local,intermediate, and global transient response curves are scaledbased on the far-field and near-field transient decay surfacecomponents respectively, computed using the trained ML decaysurface predictor, followed by linear superposition of the curvesfor each power value in the transient power profile to generatethe effective transient response curve. The runtime forgenerating thermal results for a large chip is in the order ofminutes, compared to several hours/days while using CFD/FEAbased tools with good accuracy correlation. The fast transientthermal solver is implemented on distributed ML computingplatform for parallel computation of transient thermal inferencemodel.

Thermal-constrained memory management for three-dimensional DRAM-PCM memory with deep neural network applications

For the deep neural network applications, the requirement of the memory bandwidth and volume is huge. Multiple DRAM and PCM (phase-change memory) chips are stacked to form the 3D DRAM-PCM memory, which can support different features from DRAM and PCM chips. But the thermal problem in 3D DRAM-PCM memory is more serious than the traditional memories. To constrain the peak temperature, the memory management for the 3D hybrid DRAM-PCM memory is required. In this work, we proposed the thermal-constrained memory management (TCMM) to solve this problem. TCMM applies different methods to reduce the power and constrain the peak temperature: dynamic thermal management with data preloading (DP), data management (DM), and address remapping (AR). Compared to the related works, the 3D DRAM-PCM memory using TCMM can reduce the peak temperature up to 26.25 °C and the latency up to 78.23% in our experiments. The peak temperature can be also constrained under the predefined thermal limitation.

VO2-Based Infrared Radiation Regulator with Excellent Dynamic Thermal Management Performance.

Dynamic thermal management materials attract fast-increasing interest due to their adaptability to changing environments and greater energy savings as compared to static materials. However, the high transition temperature and the low emittance tunability of the vanadium dioxide (VO2)-based infrared radiation regulators limit their practical applications. This study addresses these issues by proposing a smart infrared radiation regulator based on a Fabry-Pérot cavity structure (VO2/HfO2/Al), which is prepared by high-power impulse magnetron sputtering (HiPIMS) and has the potential for large-scale production. Remarkably, the outstanding emittance tunability reaches 0.51, and the phase transition temperature is lowered to near a room temperature of 27.5 °C by tungsten (W) doping. In addition, a numerical thermal management power of 196.3 W/m2 (at 8-14 μm band) can be obtained from 0 to 60 °C. As a proof-of-concept, the demonstrated capabilities of the VO2 infrared radiation regulator show great potentials in a wide range of applications for the thermal management of buildings and vehicles.

Effect of Phase Change Material on Dynamic Thermal Management Performance for Power Electronics Packages

Abstract High-temperature silicon carbide (SiC) die are the most critical and expensive component in electric vehicle (EV) power electronics (PE) packages and require both active and passive methods to dissipate heat during transient operation. The use of phase change materials (PCMs) to control the peak junction temperature of the SiC die and to buffer the temperature fluctuations in the package during simulated operation is modeled here. The latent heat storage potential of multiple PCM and PCM composites is explored in both single-sided and dual-sided package configurations. The results of this study show that the addition of PCM into two different styles of PE packages is an effective method for controlling the transient junction temperatures experienced during two different drive cycles. The addition of PCM in a single-sided package also serves to decrease temperature fluctuations experienced and may be used to reduce the necessary number of SiC die required for EVs, lowering the overall material cost and volume of the package by over 50%. PCM in a single-sided package may be nearly as effective as the double-sided cooling approach of a dual-sided package in the reduction of both peak junction temperature of SiC as well as controlling temperature variations between package layers.