Pith. sign in

REVIEW 4 major objections 6 minor 28 references

CPINN-ABPI: Physics-Informed Neural Networks for Accurate Power Estimation in MPSoCs

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper presents the first on-chip validation of ABPI and proposes CPINN-ABPI, a physics-informed neural network that cuts power-estimation error by 84.7% while keeping sub-millisecond inference.

desk verdict First hardware validation of ABPI is a real contribution, but the physics-informed advantage is unproven due to a supervised-vs-unsupervised confound. read the letter →

arxiv 2505.22469 v1 pith:B53WSDCU submitted 2025-05-28 cs.PF cs.LG

Mohamed R. Elshamy , Mehdi Elahi , Ahmad Patooghy , Abdel-Hameed A. Badawy This is my paper
classification cs.PFcs.LG
keywords powerestimationphysics-informedneuralnetworksmultiprocessorsystems-on-chipblindidentificationthermalmodelingNSGA-IINVIDIAJetsonAGXXavierheterogeneousSoC
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is trying to establish two things: that Alternative Blind Power Identification (ABPI), a leading method for estimating per-unit power from temperature alone, does not hold up on real hardware; and that a hybrid physics-informed neural network (CPINN-ABPI) can correct ABPI's errors well enough for practical use. The authors run the first empirical ABPI validation on a commercial MPSoC and report CPU mean absolute error falling from 3.74 W to 0.57 W and GPU from 3.74 W to 0.98 W, with weighted mean absolute percentage error dropping from 47-81% to about 12%. The result matters because unit-level power estimates are what enable predictive thermal management, dynamic voltage/frequency scaling, and thermal-attack detection on multicore chips.

What carries the argument

The engine is a two-branch estimator: a physics branch implementing ABPI's thermal update $\hat{t}=t_{\mathrm{prev}}^T A$, $\Delta t=t_{\mathrm{current}}-\hat{t}$, and $p_{\mathrm{physics}}=\Delta t\cdot(B^{-1})^T$, with the thermal matrices $A,B$ initialized from ABPI and fine-tuned during training; and a residual network branch that maps $[t_{\mathrm{prev}},t_{\mathrm{current}},P_{\mathrm{estimated}}]$ to a correction $\Delta p$. The loss $L=L_{\mathrm{data}}+\lambda_{\mathrm{phys}}L_{\mathrm{phys}}+\lambda_{\mathrm{guide}}L_{\mathrm{guide}}$ couples the branches, and NSGA-II optimizes layer count, width, activation, and the $\lambda$ weights against the two objectives of low MAE and low MAC count.

What would settle it

Hold out a set of unseen workloads on the same Jetson AGX Xavier, keep the hardware's power sensors as ground truth, and compare CPINN-ABPI, ABPI, and a plain supervised network with the same labels and no physics branch. If CPINN-ABPI's WMAPE rises well above 12% or a plain network matches its MAE, the claim that the physics-informed correction is what delivers the gain is not supported.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that ABPI's linearized thermal model, $T_r(k)=AT_r(k-1)+BP(k)$, is the accuracy bottleneck on real MPSoCs: on a Jetson AGX Xavier it yields MAE of 3.74 W for both CPU and GPU and WMAPE of 47-81%, because it collapses the heat-diffusion equation into a simplified state-space update. The paper's remedy, CPINN-ABPI, keeps ABPI's physics branch but adds a parallel residual network that learns a correction $\Delta p$, combined as $p_{\mathrm{final}}=p_{\mathrm{physics}}+\Delta p$, with a three-term loss that fits ground-truth labels while enforcing thermal consistency and staying close to the physics estimate; NSGA-II selects the network size and loss weights. On the Xavier, this brings CPU MAE to 0.57 W (84.7% reduction) and GPU MAE to 0.98 W (73.9% reduction), keeps WMAPE near 12%, and runs in 195.3 $\mu$s per inference; on a simulated six-component heterogeneous SoC the MAE improvements range from 85% to 99%.

Load-bearing premise

CPINN-ABPI is trained with per-unit ground-truth power labels, while the ABPI baseline is unsupervised; the comparison counts as a fair demonstration only if such labels are available in the settings where unit-level power estimation is needed.

Editorial extensions

If this is right

  • Unit-level power estimates on Jetson-class SoCs become accurate enough for real-time DVFS, thermal management, and thermal-Trojan detection, not just coarse server-level accounting.
  • ABPI's linear state-space thermal model is identified as the decisive source of error; correcting it with a learned residual is sufficient to cut MAE by 73.9-99% depending on the unit.
  • The approach keeps ABPI's key property: it still requires only consecutive temperature measurements and total power, not steady-state temperature.
  • The NSGA-II Pareto selection makes the accuracy-latency tradeoff explicit; the chosen Jetson model costs only 176 MACs per inference, which is why sub-millisecond operation is preserved.
  • The same recipe transfers to a simulated big.LITTLE heterogeneous SoC, where WMAPE becomes stable at low double-digit to tens of percent across all six components.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The reported gains are purchased with per-unit ground-truth power labels for training; if such labels do not exist on a target platform, the method's advantage is not established, so deployment hinges on label availability.
  • The paper compares against unsupervised ABPI; a plain supervised neural network trained on the same labels might capture much of the same accuracy, leaving open how much the physics branch specifically contributes.
  • The residual-correction recipe is general: any differentiable physics-based power/thermal estimator could be wrapped with the same three-term loss and NSGA-II tuning, so the approach may extend beyond ABPI.
  • Cross-platform transfer is tested only as initialization of $A$ and $B$ from ABPI on the same platform; whether the learned correction transfers across workloads or chip generations is not addressed.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes CPINN-ABPI, a hybrid power estimation method for MPSoCs that combines the ABPI linear thermal state-space model with a residual neural network correction, trained with a multi-component loss (data fidelity, physics consistency, and physics guidance) and tuned by NSGA-II for the accuracy/compute tradeoff. The authors report the first empirical validation of ABPI on an NVIDIA Jetson AGX Xavier board, showing that ABPI has large errors (MAE 3.74 W, WMAPE 47–81%), and claim that CPINN-ABPI reduces MAE by 84.7% (CPU) and 73.9% (GPU), with WMAPE about 12%, and similar 85–99% MAE improvements on a simulated heterogeneous SoC, while keeping inference sub-millisecond.

Significance. If the central claim were fully established, the paper would make a useful empirical contribution: it would be the first public validation of ABPI on commercial hardware, and it would demonstrate that a supervised residual network trained with a physics-inspired loss can substantially improve blind power identification. The paper also contributes a new Jetson dataset and an NSGA-II-based architecture search for real-time power models. However, the experimental design as presented does not separate the effect of supervision from the effect of the physics-informed branch, and the physics constraint itself is not shown to be an independent prior because the thermal matrices are re-estimated during training. The reported gains are therefore not yet attributable to the proposed method's distinctive features, and a plain supervised baseline and a proper ablation are needed before the significance claimed in the title and abstract can be accepted.

major comments (4)
  1. [II-B, Algorithm 1, Table II] The headline accuracy gains are confounded by supervision: CPINN-ABPI is trained with per-unit ground-truth labels (L_data in Eq. 3, Algorithm 1 lines 14–17), whereas the ABPI baseline is blind and unsupervised, using only temperature and total power. On this evidence alone, the 84.7% CPU and 73.9% GPU MAE reductions in Table II could be achieved by any supervised residual network and do not establish the contribution of the physics branch, Eq. 4, or Eq. 5. The paper should add a plain supervised baseline (e.g., an MLP with the same inputs and training protocol) and an ablation with λ_phys = λ_guide = 0.
  2. [II-A, Eq. 4, Algorithm 1] The physics consistency loss is not an independent constraint: the matrices A' and B' in Eq. 4 are initialized from ABPI estimates on the same data and then further optimized by gradient descent (Algorithm 1 lines 21–22). Consequently L_phys can be minimized by changing A' and B' to conform to the network's power predictions, so it does not independently enforce thermodynamic consistency. To support the 'physics-informed' claim, the authors should either fix A and B to independently identified values, or demonstrate that the trained A', B' remain close to physically meaningful values and that the accuracy gain is not lost when L_phys is replaced by a purely data-driven regularization.
  3. [Algorithm 2, Section IV-A] The NSGA-II selection in Algorithm 2 evaluates fitness on D_test (line 9) and selects the Pareto-optimal architecture using that test set. Reporting the selected model's performance on the same D_test in Section IV therefore gives optimistically biased estimates. The 10-fold cross-validation applied after selection does not remove this selection bias because the architecture was chosen using the test data. The evaluation protocol should use a separate held-out test set that is never touched during NSGA-II, or use nested cross-validation.
  4. [Table II, Section IV-B] Table II reports identical ABPI MAE (3.74 W) and MSE (32.10 W^2) for both CPU and GPU, which is implausible for two units with different power profiles and workloads. This suggests a transcription or computation error in the baseline results; because the improvement percentages are computed relative to these ABPI numbers, the baseline must be verified and corrected.
minor comments (6)
  1. [Figure 1] The caption reads 'Experimental setup overflow' instead of 'overview'.
  2. [Eq. 5] The heading 'Physics Guidance Loss::' contains a double colon; the extra colon should be removed.
  3. [Section II-A] The notation for the physics branch is inconsistent: the text uses p_physics = Δt·(B^{-1})^T and t_prev^T A, while Algorithm 1 writes P_physics ← ΔT·(B'^{-1})^T and A'^T T_prev; please unify the transpose conventions and clarify the shapes of A, B, and the vectors.
  4. [Section III] The paper does not report the number of samples in the training, validation, and test sets, the duration of each workload trace, or the train/test split ratio; these details are needed to interpret the reported 10-fold cross-validation and the final test errors.
  5. [Section IV-A] The NSGA-II description gives the number of generations and population size, but not the crossover and mutation rates or the ranges of the searched hyperparameters, which limits reproducibility.
  6. [Section IV-B] Figure 8 shows a 2500-second test dataset, but the paper does not state whether this is a single continuous trace or a concatenation of the ten workloads in Table I, nor how many independent runs were averaged; please clarify.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the reported gains are empirical supervised-vs-unsupervised measurements; the data-fitted physics matrices weaken the 'physics-informed' interpretation but do not make the result equivalent to its inputs.

full rationale

The paper's derivation chain is an empirical evaluation, not a formal proof. CPINN-ABPI is trained on ground-truth unit power via the data-fidelity loss (Eq. 3, Algorithm 1) and then compared against the blind, unsupervised ABPI baseline; the MAE/WMAPE reductions are measured on held-out data (10-fold cross-validation and a 2500-second test set), so the headline accuracy result is not defined into existence. The physics branch initializes A and B from ABPI's fitted thermal model and fine-tunes them during training, so Eq. 4's physics-consistency term enforces consistency with a data-derived linear surrogate rather than an external first-principles law. That is a real threat to the claim that the accuracy gain validates 'physics-informed' learning specifically, because the supervised data-fidelity loss already has access to p_true while ABPI does not. But this is a comparison confound, not a circular reduction in which a prediction is equivalent to its inputs by construction. No load-bearing self-citation appears: the authors' own references [4] and [10] are contextual only, and no uniqueness theorem or ansatz is imported via self-citation. The duplicated ABPI CPU/GPU metrics in Table II are anomalous and worth checking, but they do not constitute circularity. Overall, the central claim is self-contained as an empirical comparison, so the circularity score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical entities are postulated. The physics in CPINN-ABPI comes from the fitted linear ABPI model plus a trainable residual network; the paper introduces no new particles, forces, dimensions, or conserved quantities.

free parameters (5)
  • Thermal matrix A = Estimated by ABPI from each platform's temperature and total-power traces; then fine-tuned during CPINN training
    The physics branch (Section II-A, Algorithm 1) treats A as a learned matrix, not a theoretically derived quantity.
  • Thermal matrix B = Estimated by ABPI; then fine-tuned during CPINN training
    Same as A; B maps power to temperature changes and is fitted to data.
  • lambda_phys = Not reported; selected by NSGA-II
    Weight of the physics consistency loss in Equation 6, chosen by the genetic algorithm on validation data.
  • lambda_guide = Not reported; selected by NSGA-II
    Weight of the physics guidance loss in Equation 6, chosen by NSGA-II.
  • Residual network architecture = Jetson: 1 layer with 21 neurons; simulated SoC: 2 layers with 80 and 55 neurons
    Selected by NSGA-II (Section IV-A) to balance MAE and MAC count, then validated with 10-fold cross-validation.
assumptions (4)
  • domain assumption The linear state-space thermal model Tr(k) = A Tr(k-1) + B P(k) is sufficient as the physics prior for power estimation.
    Equation 2 and the physics branch in Algorithm 1 depend on this linearization; ABPI's known inaccuracy (Section I) shows the assumption is questionable, and CPINN still uses it as its physical constraint in Equation 4.
  • domain assumption Per-unit ground-truth power values are available for training the residual network.
    The data fidelity loss (Equation 3) and Algorithm 1 require p_true for each unit, yet the motivating problem states that unit-level power sensors are impractical (Section I).
  • domain assumption A 1-second sampling interval captures the thermal dynamics relevant to power estimation.
    All experiments use 1-second intervals (Section III); faster transients are unmodeled.
  • domain assumption HotSpot v7 and CoMeT simulations faithfully reproduce the thermal and power behavior of the heterogeneous SoC.
    Section III-B uses HotSpot for temperature and CoMeT for ground-truth power; errors in either simulator propagate to the reported gains.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CPINN-ABPI: Physics-Informed Neural Networks for Accurate Power Estimation in MPSoCs." pith.science (2026). https://pith.science/paper/B53WSDCU

@misc{pith2026250522469,
  author       = {Pith},
  title        = {Pith review of: CPINN-ABPI: Physics-Informed Neural Networks for Accurate Power Estimation in MPSoCs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B53WSDCU}},
  note         = {Machine review of arXiv:2505.22469}
}
abstract

Efficient thermal and power management in modern multiprocessor systems-on-chip (MPSoCs) demands accurate power consumption estimation. One of the state-of-the-art approaches, Alternative Blind Power Identification (ABPI), theoretically eliminates the dependence on steady-state temperatures, addressing a major shortcoming of previous approaches. However, ABPI performance has remained unverified in actual hardware implementations. In this study, we conduct the first empirical validation of ABPI on commercial hardware using the NVIDIA Jetson Xavier AGX platform. Our findings reveal that, while ABPI provides computational efficiency and independence from steady-state temperature, it exhibits considerable accuracy deficiencies in real-world scenarios. To overcome these limitations, we introduce a novel approach that integrates Custom Physics-Informed Neural Networks (CPINNs) with the underlying thermal model of ABPI. Our approach employs a specialized loss function that harmonizes physical principles with data-driven learning, complemented by multi-objective genetic algorithm optimization to balance estimation accuracy and computational cost. In experimental validation, CPINN-ABPI achieves a reduction of 84.7\% CPU and 73.9\% GPU in the mean absolute error (MAE) relative to ABPI, with the weighted mean absolute percentage error (WMAPE) improving from 47\%--81\% to $\sim$12\%. The method maintains real-time performance with 195.3~$\mu$s of inference time, with similar 85\%--99\% accuracy gains across heterogeneous SoCs.

Figures

Figures reproduced from arXiv: 2505.22469 by the authors.

Figure 3
Figure 3. Jetson Xavier AGX with CPU and GPU Power and Temperature [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 1
Figure 1. Experimental setup overflow for the simulation and the practical [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 5
Figure 5. Custom Dataset Generation Workflow for a Heterogeneous SoC [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figures from the paper (4 more)
Figure 6
Figure 6. Figure 6: Pareto front showing the tradeoff between prediction accuracy (MAE in Watts) and computational cost (MAC operations) for various CPINN-ABPI [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Average training and validation loss curves during 10-fold cross [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Comparison of actual power measurements versus predictions from ABPI and CPINN-ABPI for the CPU and GPU of the Jetson AGX Xavier under [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Power prediction comparison between CPINN-ABPI and ABPI. (Left) Little CPU unit performance over 2000 seconds. (Right) Big Core2 performance [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

28 extracted references · 24 canonical work pages

  1. [1]

    Teem: Online thermal- and energy-efficiency management on cpu-gpu mpsocs,

    S. Isuwa, S. Dey, A. K. Singh, and K. McDonald-Maier, “Teem: Online thermal- and energy-efficiency management on cpu-gpu mpsocs,” in 2019 Design, Automation & Test in Europe Conference & Exhibition (DATE), 2019, pp. 438–443

    work page 2019

  2. [2]

    An evaluation framework for dynamic thermal management strategies in 3d multiprocessor system-on- chip co-design,

    D. Huang, L. Costero, and D. Atienza, “An evaluation framework for dynamic thermal management strategies in 3d multiprocessor system-on- chip co-design,”IEEE Transactions on Parallel and Distributed Systems, vol. 35, no. 11, pp. 2161–2176, 2024

    work page 2024

  3. [3]

    Blind identification of power sources in processors,

    S. Reda and A. Belouchrani, “Blind identification of power sources in processors,” inDesign, Automation & Test in Europe Conference & Exhibition (DATE), 2017, 2017, pp. 1739–1744

    work page 2017

  4. [4]

    Cluster- bpi: Efficient fine-grain blind power identification for defending against hardware thermal trojans in multicore socs,

    M. R. Elshamy, M. Elahi, A. Patooghy, and A.-H. A. Badawy, “Cluster- bpi: Efficient fine-grain blind power identification for defending against hardware thermal trojans in multicore socs,” in2024 IEEE International Performance, Computing, and Communications Conference (IPCCC), 2024, pp. 1–6

    work page 2024

  5. [5]

    Blind identification of thermal models and power sources from thermal measurements,

    S. Reda, K. Dev, and A. Belouchrani, “Blind identification of thermal models and power sources from thermal measurements,”IEEE Sensors J., 2018

    work page 2018

  6. [6]

    (2023) Snapdragon 8 gen 3 mobile hardware development kit (hdk-8650)

    Qualcomm Technologies, Inc. (2023) Snapdragon 8 gen 3 mobile hardware development kit (hdk-8650). Distributed by Lantronix eStore. [Online]. Available: https://estore.lantronix.com/products/ copy-of-snapdragon-8-gen-3-mobile-hardware-development-kit-hdk-8650

    work page 2023

  7. [7]

    Emulation of an asic power, temperature and aging monitor system for fpga prototyping,

    A. Listl, D. Mueller-Gritschneder, F. Kluge, and U. Schlichtmann, “Emulation of an asic power, temperature and aging monitor system for fpga prototyping,” in2018 IEEE 24th International Symposium on On-Line Testing And Robust System Design (IOLTS), 2018, pp. 220–225

    work page 2018

  8. [8]

    Performance-driven dynamic thermal management of mpsoc based on task rescheduling,

    K. Ganeshpure and S. Kundu, “Performance-driven dynamic thermal management of mpsoc based on task rescheduling,”ACM Trans. Des. Autom. Electron. Syst., vol. 19, no. 2, Mar. 2014. [Online]. Available: https://doi.org/10.1145/2566661

    work page doi:10.1145/2566661 2014

Show all 28 references
  1. [9]

    Bic: Blind identification countermeasure for malicious thermal sensor attacks in mobile socs,

    M. Abdelrehimet al., “Bic: Blind identification countermeasure for malicious thermal sensor attacks in mobile socs,” inISQED, 2022

    2022

  2. [10]

    Matter: Multi-stage adaptive thermal trojan for efficiency & resilience degradation,

    M. Elahi, M. R. Elshamy, A.-H. Badawy, M. Fazeli, and A. Patooghy, “Matter: Multi-stage adaptive thermal trojan for efficiency & resilience degradation,” 2024. [Online]. Available: https://arxiv.org/abs/2412.00226

    2024 arXiv

  3. [11]

    A systematic methodology to generate decomposable and responsive power models for cmps,

    R. Bertran, M. Gonzalez, X. Martorell, N. Navarro, and E. Ayguade, “A systematic methodology to generate decomposable and responsive power models for cmps,”IEEE Transactions on Computers, vol. 62, no. 7, pp. 1289–1302, 2013. Fig. 8. Comparison of actual power measurements vers...

    2013

  4. [12]

    Complete system power estimation using processor performance events,

    W. L. Bircher and L. K. John, “Complete system power estimation using processor performance events,”IEEE Transactions on Computers, vol. 61, no. 4, pp. 563–577, 2012

    2012

  5. [13]

    Evaluating the effectiveness of model-based power characterization,

    J. C. McCullough, Y . Agarwal, J. Chandrashekar, S. Kuppuswamy, A. C. Snoeren, and R. K. Gupta, “Evaluating the effectiveness of model-based power characterization,” inProceedings of the 2011 USENIX Conference on USENIX Annual Technical Conference, ser. USENIXATC’11. USA: USEN...

    2011

  6. [14]

    Towards a generic power estimator,

    L. F. Cupertino, G. Da Costa, and J.-M. Pierson, “Towards a generic power estimator,”Computer Science - Research and Development, vol. 30, no. 2, pp. 145–153, 2015. [Online]. Available: https://doi.org/10.1007/s00450-014-0264-x

    2015 doi

  7. [15]

    An artificial neural network approach to power consumption model construction for servers in cloud data centers,

    W. Lin, G. Wu, X. Wang, and K. Li, “An artificial neural network approach to power consumption model construction for servers in cloud data centers,”IEEE Transactions on Sustainable Computing, vol. 5, no. 3, pp. 329–340, 2020

    2020

  8. [16]

    Fine-grained power modeling of multicore processors using FFNNs,

    M. Sagi, N. A. Vu Doan, N. Fasfous, T. Wild, and A. Herkersdorf, “Fine-grained power modeling of multicore processors using FFNNs,” International Journal of Parallel Programming, vol. 50, no. 2, pp. 243–266, April 2022. [Online]. Available: https://doi.org/10.1007/ s10766-022-00730-9

    2022

  9. [17]

    Alternating blind identification of power sources for mobile socs,

    S. Chetoui, M. Chen, A. Golas, F. Hijaz, A. Belouchrani, and S. Reda, “Alternating blind identification of power sources for mobile socs,” inProceedings of the 2022 ACM/SPEC on International Conference on Performance Engineering, ser. ICPE ’22. New York, NY , USA: Association ...

    2022

  10. [18]

    Understanding the sources of power consumption in mobile socs,

    M. Said, S. Chetoui, A. Belouchrani, and S. Reda, “Understanding the sources of power consumption in mobile socs,” in2018 Ninth International Green and Sustainable Computing Conference (IGSC), 2018, pp. 1–7

    2018

  11. [19]

    Nvidia jetson agx xavier: Embedded system for autonomous machines,

    NVIDIA Corporation, “Nvidia jetson agx xavier: Embedded system for autonomous machines,” 2025. [Online]. Available: https://www.nvidia. com/en-us/autonomous-machines/embedded-systems/jetson-agx-xavier/

    2025

  12. [20]

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,

    M. Raissi, P. Perdikaris, and G. Karniadakis, “Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,” Journal of Computational Physics, vol. 378, pp. 686–707, 2019. [Online]. Ava...

    2019

  13. [21]

    A fast and elitist multiobjective genetic algorithm: Nsga-ii,

    K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: Nsga-ii,”IEEE Transactions on Evo- lutionary Computation, vol. 6, no. 2, pp. 182–197, 2002

    2002

  14. [22]

    Nsga-pinn: A multi-objective optimization method for physics-informed neural network training,

    B. Lu, C. Moya, and G. Lin, “Nsga-pinn: A multi-objective optimization method for physics-informed neural network training,” Algorithms, vol. 16, no. 4, 2023. [Online]. Available: https: //www.mdpi.com/1999-4893/16/4/194

    2023

  15. [23]

    M. N. ¨Ozisik,Heat Conduction. Wiley, 1993

    1993

  16. [24]

    stress-ng,

    C. I. King, “stress-ng,” 2023, a tool to load and stress a computer system. [Online]. Available: https://github.com/ColinIanKing/stress-ng

    2023

  17. [25]

    Jetson stats tool (jtop),

    NVIDIA, “Jetson stats tool (jtop),” 2025, accessed: 2025-01-04. [Online]. Available: https://developer.nvidia.com/embedded/community/ jetson-projects/jetson stats

    2025

  18. [26]

    Thermal modeling and validation of a real-world mobile ap,

    Y .-H. Gong, J. J. Yoo, and S. W. Chung, “Thermal modeling and validation of a real-world mobile ap,”IEEE Design & Test, vol. 35, no. 1, pp. 55–62, 2018

    2018

  19. [27]

    Hotspot: A temperature modeling tool for ics,

    K. Skadron, M. R. Stan, W. Huang, S. Velusamy, and K. Sankara- narayanan, “Hotspot: A temperature modeling tool for ics,” https: //github.com/uvahotspot/HotSpot, 2024, accessed: 2024-07-22

    2024

  20. [28]

    Comet: An integrated interval thermal simulation toolchain for 2d, 2.5d, and 3d processor-memory systems,

    L. Siddhu, R. Kedia, S. Pandey, M. Rapp, A. Pathania, J. Henkel, and P. R. Panda, “Comet: An integrated interval thermal simulation toolchain for 2d, 2.5d, and 3d processor-memory systems,”ACM Trans. Archit. Code Optim., vol. 19, no. 3, Aug. 2022. [Online]. Available: https://...

    2022 doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.