Pith. sign in

REVIEW 4 major objections 3 minor 54 references

DiffAxE: Diffusion-driven Hardware Accelerator Generation and Design Space Exploration

T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A diffusion-based generative model can learn hardware-performance mappings in an O(10^17)-scale accelerator design space and generate designs conditioned on target performance, beating Bayesian optimization in error and speed.

desk verdict Conditional diffusion for accelerator DSE is a genuinely new angle, but the abstract's headline numbers are not yet supported by the evidence shown. read the letter →

arxiv 2508.10303 v2 pith:2MTWDFF5 submitted 2025-08-14 cs.AR

Arkapravo Ghosh , Abhishek Moitra , Abhiroop Bhattacharjee , Ruokai Yin , Priyadarshini Panda This is my paper
classification cs.AR
keywords designspaceexplorationdiffusionmodelshardwareacceleratorgenerationconditionalBayesianoptimizationLLMinferenceASICFPGA
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 tries to establish that a diffusion model, originally built for image synthesis, can be repurposed to learn the non-differentiable, many-to-one mapping between hardware design configurations and performance metrics. The key move is to treat a hardware design as a 1-D image and condition the diffusion process on a target performance, enabling direct generation of accelerator designs instead of slow iterative search. The authors show that this approach achieves 0.86% lower generation error than Bayesian optimization with a 17000x speedup, and also outperforms existing generative DSE methods on design spaces with roughly $10^{17}$ configurations. If correct, this would make design space exploration for AI-oriented accelerators dramatically faster, including for LLM inference, where the method reports 3.37x and 7.75x lower energy-delay product on ASIC and FPGA compared to the DOSA framework.

What carries the argument

The central object is a conditional diffusion model applied to hardware design: each accelerator configuration is encoded as a 1-D sequence (the 'image'), and a denoising process is trained to reverse a corruption process while being conditioned on target performance values. This machinery carries the argument because it learns the inverse of the performance-evaluation function even when that function is non-differentiable, non-injective, and defined on an irregular O($10^{17}$) discrete space.

What would settle it

Run a held-out validation: train the diffusion generator on a random subset of an accelerator design space, then condition it on performance targets drawn from untouched configurations and check against cycle-accurate simulation; the central claim fails if the generated designs meet target performance only for configurations close to training samples.

Watch

Extended reading notes

Core claim

The central claim is that hardware design space exploration can be reframed as conditional generation: a design configuration is treated as a 1-D image, and a diffusion model learns the distribution of designs that correspond to a user-specified target performance. This formulation lets the model capture non-differentiable and non-bijective hardware-performance mappings without needing gradients or iterative sampling. The paper reports that this diffusion-driven generator achieves 0.86% lower generation error than Bayesian optimization with a 17000x speedup, and 30% lower error than GANDSE at 1.83x slower search. In a structured DSE setting, it attains 9.8% lower energy-delay product and 6%

Load-bearing premise

The central premise is that the finite set of design-performance pairs used to train the model fairly represents the entire O($10^{17}$) design space; if those samples are sparse or biased, generated designs may match target metrics in training but miss them on real hardware.

Editorial extensions

If this is right

  • Design space exploration becomes a single trained generation step rather than an iterative search, reducing search time by orders of magnitude (17,000x versus Bayesian optimization in the paper's comparison).
  • On O(10^17) design spaces, the conditional generator lowers energy-delay product by 9.8% and raises performance by 6% while speeding up search by up to 145.6x and 1312x versus existing optimization methods.
  • For LLM inference accelerators, the approach achieves 3.37x lower energy-delay product on a 32nm ASIC and 7.75x lower on a Xilinx Ultrascale+ VPU13 FPGA compared to the state-of-the-art DOSA framework.
  • The generative formulation handles many-to-one, non-differentiable hardware-performance mappings without gradient information, which is exactly the regime where Bayesian optimization, gradient descent, and reinforcement learning struggle.

Reading between the lines

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

  • The same conditional diffusion formulation could be applied to other discrete combinatorial design problems with expensive black-box evaluations, such as compiler flag selection, network pruning, or hardware-software co-design, wherever a dataset of configuration-metric pairs exists.
  • The reported 17000x speedup is at generation time; a fair end-to-end comparison would also count the one-time cost of building the training dataset and training the diffusion model, which the abstract does not break out.
  • Because the model conditions on target performance analogously to how image diffusion conditions on captions, adding more objectives such as latency, area, and energy should be as simple as appending conditioning channels, potentially extending the method to multi-objective DSE without new machinery.
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 / 3 minor

Summary. The paper proposes DiffAxE, a diffusion-based generative framework for hardware accelerator design space exploration. It frames accelerator configuration generation as 1-D image synthesis conditioned on a target performance metric, aiming to learn non-differentiable, non-bijective mappings from design configurations to performance. The abstract reports strong empirical results: 0.86% lower generation error than Bayesian optimization at 17000x speedup, 30% lower error than GANDSE at 1.83x slower search, improved energy-delay product (EDP) and performance in a structured DSE setting with large speedups, and 3.37x/7.75x lower EDP for LLM inference on ASIC/FPGA compared to DOSA. The central claim is that unconditional generative modeling avoids iterative sampling and handles O(10^17) design spaces better than previous optimization and classification-based methods.

Significance. If the reported results hold under rigorous verification, the contribution is significant: it would demonstrate that conditional diffusion models can navigate irregular, many-to-one hardware design spaces more effectively than established DSE methods, and would open a generative-design direction for accelerator exploration. The framing addresses a real limitation of iterative optimization and of classification-based DSE, and the external baselines (BO, GANDSE, DOSA) provide a meaningful comparative context. However, the significance is conditional: the abstract alone does not establish that the generated designs are valid or that the performance gains are measured independently of the surrogate used to train and evaluate the model. The paper currently presents a promising empirical claim rather than a verified result, and the absence of uncertainty quantification and experimental protocol details limits the ability to assess its robustness.

major comments (4)
  1. [Abstract, '0.86% lower generation error'] The metric 'generation error' is not defined. Is it computed with a cycle-accurate simulator, RTL synthesis, or the same surrogate performance model used to produce training labels? If the latter, the reported 17000x speedup and 0.86% improvement may reflect fitting to the surrogate rather than generating implementable accelerators. Please define the error metric, the measurement pipeline, and report confidence intervals or multiple-seed results, since 0.86% is small relative to typical run-to-run variation.
  2. [Abstract, 'non-differentiable, non-bijective hardware-performance mappings'] The central claim requires evidence that generated designs are valid and achieve the conditioned performance. In an O(10^17) space with sparse labeled data, a conditional diffusion model can produce configurations that interpolate the training manifold but violate microarchitectural constraints, timing, or area limits. Please report post-generation validation: the fraction of designs that pass structural/constraint checks, the performance achieved in cycle-accurate simulation or synthesis, and any comparison against the training distribution (e.g., distance to nearest training design).
  3. [Abstract, 'outperforms GANDSE' and 'DOSA' comparisons] The comparisons to GANDSE and DOSA report error, EDP, and speedups without experimental protocol. Specify the number of independent runs, random seeds, initial samples, hyperparameter settings, and whether the reported speedups include offline training/amortization time or only search iterations. The '145.6x and 1312x faster search' numbers are not interpretable without a clear timing definition.
  4. [Abstract, O(10^17) design space and training coverage] The abstract does not describe the training dataset size, sampling strategy, or coverage of the design space. If the labeled design-performance pairs are sparse or biased, the conditional generator will only appear to work within the sampled region, and the headline results may not generalize. Please report the dataset construction, the number of design points, and whether the evaluation design points are held out from training.
minor comments (3)
  1. [Abstract] Abbreviations such as DSE, EDP, DOSA, and GANDSE are used without definition at first mention; please expand them.
  2. [Abstract, '1-D image synthesis'] The analogy to 1-D image synthesis is intriguing but underspecified. Please clarify how a hardware configuration is encoded as a 1-D image, how conditioning on target performance is applied, and what the denoising objective corresponds to in the hardware design context.
  3. [Abstract, numerical results] Consider reporting confidence intervals or error bars for the headline 0.86% and 30% figures, and state the number of test design points used for each reported comparison.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found in the available abstract; evaluation is against external baselines and no load-bearing self-citation or definitional reduction is evident.

full rationale

The abstract describes a conditional generative model trained on hardware design-performance pairs and evaluates it against external baselines (Bayesian optimization, GANDSE, DOSA). This gives the central claim independent grounding: the trained model is not a fitted constant renamed as a prediction, and the reported comparisons concern generation error, EDP, and search speed relative to other methods. The abstract does not define 'generation error' precisely, nor does it state that the metric is computed from the same surrogate used to generate training labels, so no specific reduction can be exhibited. The flagged concern about training coverage and lack of physical/cycle-accurate verification is a generalization and validity risk, not a circularity argument. No self-citation, uniqueness import, ansatz-smuggling, or renaming pattern is visible in the provided manuscript text. Therefore, under the hard rule that circularity must be demonstrated by quote and explicit reduction, the finding is no significant circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

Only abstract was available for review. The ledger entries reflect assumptions visible from the abstract; full methodology may contain additional parameters and assumptions.

free parameters (3)
  • Diffusion model weights (including transformer/UNet parameters)
    Learned from a dataset of design-performance pairs; the central claim depends on the model learning a conditional mapping.
  • Diffusion sampling hyperparameters (e.g., denoising steps, noise schedule)
    Chosen by authors; affect generation quality and search speed, but not explicit in abstract.
  • Conditioning scale
    In conditional diffusion, the weight of the target performance condition affects output; not detailed in abstract.
assumptions (4)
  • ad hoc to paper The design-performance mapping can be represented as a 1-D image synthesis task.
    Core modeling choice in the abstract; not derived from first principles.
  • domain assumption Performance metrics used for conditioning (e.g., EDP, accuracy) are reliable and computable for all design configurations.
    The method assumes these metrics are available from simulation or measurement; if not, the conditioned generation is meaningless.
  • domain assumption The training dataset is representative of the full O(10^17) design space.
    Generalization of the conditional model depends on this; explicitly discussed in weakest_assumption.
  • standard math Diffusion model sampling can efficiently produce diverse, high-quality designs.
    Relies on established diffusion generative model properties, though these do not guarantee hardware design validity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DiffAxE: Diffusion-driven Hardware Accelerator Generation and Design Space Exploration." pith.science (2026). https://pith.science/paper/2MTWDFF5

@misc{pith2026250810303,
  author       = {Pith},
  title        = {Pith review of: DiffAxE: Diffusion-driven Hardware Accelerator Generation and Design Space Exploration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2MTWDFF5}},
  note         = {Machine review of arXiv:2508.10303}
}
read the original abstract

Design space exploration (DSE) is critical for developing optimized hardware architectures, especially for AI workloads such as deep neural networks (DNNs) and large language models (LLMs), which require specialized acceleration. As model complexity grows, accelerator design spaces have expanded to O(10^17), becoming highly irregular, non-convex, and exhibiting many-to-one mappings from design configurations to performance metrics. This complexity renders direct inverse derivation infeasible and necessitates heuristic or sampling-based optimization. Conventional methods - including Bayesian optimization, gradient descent, reinforcement learning, and genetic algorithms - depend on iterative sampling, resulting in long runtimes and sensitivity to initialization. Deep learning-based approaches have reframed DSE as classification using recommendation models, but remain limited to small-scale (O(10^3)), less complex design spaces. To overcome these constraints, we propose a generative approach that models hardware design as 1-D image synthesis conditioned on target performance, enabling efficient learning of non-differentiable, non-bijective hardware-performance mappings. Our framework achieves 0.86% lower generation error than Bayesian optimization with a 17000x speedup, and outperforms GANDSE with 30% lower error at only 1.83x slower search. We further extend the method to a structured DSE setting, attaining 9.8% lower energy-delay product (EDP) and 6% higher performance, with up to 145.6x and 1312x faster search compared to existing optimization methods on O(10^17) design spaces. For LLM inference, our method achieves 3.37x and 7.75x lower EDP on a 32nm ASIC and Xilinx Ultrascale+ VPU13 FPGA, respectively, compared to the state-of-the-art DOSA framework.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

54 extracted references · 47 canonical work pages

  1. [1]

    Zhao et al., ``Neural processing units for on-device machine learning: Opportunities and challenges,'' in Proceedings of the 60th Annual Design Automation Conference (DAC), 2023

    T. Zhao et al., ``Neural processing units for on-device machine learning: Opportunities and challenges,'' in Proceedings of the 60th Annual Design Automation Conference (DAC), 2023

    work page 2023

  2. [2]

    Sze, Y.-H

    V. Sze, Y.-H. Chen, T.-J. Yang, and J. S. Emer, ``Efficient processing of deep neural networks: A tutorial and survey,'' Proceedings of the IEEE, vol. 105, no. 12, pp. 2295--2329, 2017

    work page 2017

  3. [3]

    Reagen et al., ``Weightless: Lossy weight encoding for deep neural network compression,'' in Proceedings of the 58th ACM/IEEE Design Automation Conference (DAC), 2021

    B. Reagen et al., ``Weightless: Lossy weight encoding for deep neural network compression,'' in Proceedings of the 58th ACM/IEEE Design Automation Conference (DAC), 2021

    work page 2021

  4. [4]

    Chen et al., ``Eyeriss: An energy-efficient reconfigurable accelerator for deep convolutional neural networks,'' in IEEE International Solid-State Circuits Conference (ISSCC)

    Y.-H. Chen et al., ``Eyeriss: An energy-efficient reconfigurable accelerator for deep convolutional neural networks,'' in IEEE International Solid-State Circuits Conference (ISSCC). 1em plus 0.5em minus 0.4em IEEE, 2016, pp. 262--263

    work page 2016

  5. [5]

    Y. Shao et al., ``Simba: Scaling deep-learning inference with multi-chip-module-based architecture,'' in Proceedings of the 52nd Annual IEEE/ACM International Symposium on Microarchitecture, 2019, pp. 14--27

    work page 2019

  6. [6]

    Q. Huang et al., ``Learning a continuous and reconstructible latent space for hardware accelerator design,'' in 2022 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS). 1em plus 0.5em minus 0.4em IEEE, 2022, pp. 277--287

    work page 2022

  7. [7]

    Touvron, M

    H. Touvron, M. Cord, M. Douze, F. Massa, A. Sablayrolles, and H. J \'e gou, ``Training data-efficient image transformers & distillation through attention,'' in International conference on machine learning. 1em plus 0.5em minus 0.4em PMLR, 2021, pp. 10\,347--10\,357

    work page 2021

  8. [8]

    C. Hong et al., ``Dosa: Differentiable model-based one-loop search for dnn accelerators,'' in Proceedings of the 56th Annual IEEE/ACM International Symposium on Microarchitecture, 2023, pp. 209--224

    work page 2023

Show all 54 references
  1. [9]

    Liu et al., ``Darts: Differentiable architecture search,'' arXiv preprint arXiv:1806.09055, 2018

    H. Liu et al., ``Darts: Differentiable architecture search,'' arXiv preprint arXiv:1806.09055, 2018

    2018 arXiv

  2. [10]

    B. Wu et al., ``Fbnet: Hardware-aware efficient convnet design via differentiable neural architecture search,'' in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 10\,734--10\,742

    2019

  3. [11]

    Parashar et al., ``Timeloop: A systematic approach to dnn accelerator evaluation,'' in 2019 IEEE international symposium on performance analysis of systems and software (ISPASS)

    A. Parashar et al., ``Timeloop: A systematic approach to dnn accelerator evaluation,'' in 2019 IEEE international symposium on performance analysis of systems and software (ISPASS). 1em plus 0.5em minus 0.4em IEEE, 2019, pp. 304--315

    2019

  4. [12]

    Y. N. Wu et al., ``Accelergy: An architecture-level energy estimation methodology for accelerator designs,'' in 2019 IEEE/ACM International Conference on Computer-Aided Design (ICCAD). 1em plus 0.5em minus 0.4em IEEE, 2019, pp. 1--8

    2019

  5. [13]

    Samajdar et al., ``Scale-sim: Systolic cnn accelerator simulator,'' arXiv preprint arXiv:1811.02883, 2018

    A. Samajdar et al., ``Scale-sim: Systolic cnn accelerator simulator,'' arXiv preprint arXiv:1811.02883, 2018

    2018 arXiv

  6. [14]

    S.-C. Kao et al., ``Confuciux: Autonomous hardware resource assignment for dnn accelerators using reinforcement learning,'' in 2020 53rd Annual IEEE/ACM International Symposium on Microarchitecture (MICRO). 1em plus 0.5em minus 0.4em IEEE, 2020, pp. 622--636

    2020

  7. [15]

    ------, ``Gamma: Automating the hw mapping of dnn models on accelerators via genetic algorithm,'' in Proceedings of the 39th International Conference on Computer-Aided Design, 2020, pp. 1--9

    2020

  8. [16]

    1em plus 0.5em minus 0.4em IEEE, 2022, pp

    ------, ``Digamma: Domain-aware genetic algorithm for hw-mapping co-optimization for dnn accelerators,'' in 2022 Design, Automation & Test in Europe Conference & Exhibition (DATE). 1em plus 0.5em minus 0.4em IEEE, 2022, pp. 232--237

    2022

  9. [17]

    Lin et al., ``Naas: Neural accelerator architecture search,'' in 2021 58th ACM/IEEE Design Automation Conference (DAC)

    Y. Lin et al., ``Naas: Neural accelerator architecture search,'' in 2021 58th ACM/IEEE Design Automation Conference (DAC). 1em plus 0.5em minus 0.4em IEEE, 2021, pp. 1051--1056

    2021

  10. [18]

    Reagen, J

    B. Reagen, J. M. Hern \'a ndez-Lobato, R. Adolf, M. Gelbart, P. Whatmough, G.-Y. Wei, and D. Brooks, ``A case for efficient accelerator design space exploration via bayesian optimization,'' in 2017 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED). ...

    2017

  11. [19]

    Sakhuja et al., ``Polaris: Multi-fidelity design space exploration of deep learning accelerators,'' arXiv preprint arXiv:2412.15548, 2024

    C. Sakhuja et al., ``Polaris: Multi-fidelity design space exploration of deep learning accelerators,'' arXiv preprint arXiv:2412.15548, 2024

    2024 arXiv

  12. [20]

    Seo et al., ``Airchitect v2: Learning the hardware accelerator design space through unified representations,'' arXiv preprint arXiv:2501.09954, 2025

    J. Seo et al., ``Airchitect v2: Learning the hardware accelerator design space through unified representations,'' arXiv preprint arXiv:2501.09954, 2025

    2025 arXiv

  13. [21]

    Samajdar et al., ``Airchitect: Learning custom architecture design and mapping space,'' arXiv preprint arXiv:2108.08295, 2021

    A. Samajdar et al., ``Airchitect: Learning custom architecture design and mapping space,'' arXiv preprint arXiv:2108.08295, 2021

    2021 arXiv

  14. [22]

    Blocklove et al., ``Chip-chat: Challenges and opportunities in conversational hardware design,'' in 2023 ACM/IEEE 5th Workshop on Machine Learning for CAD (MLCAD)

    J. Blocklove et al., ``Chip-chat: Challenges and opportunities in conversational hardware design,'' in 2023 ACM/IEEE 5th Workshop on Machine Learning for CAD (MLCAD). 1em plus 0.5em minus 0.4em IEEE, 2023, pp. 1--6

    2023

  15. [23]

    Y. Fu et al., ``Gpt4aigchip: Towards next-generation ai accelerator design automation via large language models,'' in 2023 IEEE/ACM International Conference on Computer Aided Design (ICCAD). 1em plus 0.5em minus 0.4em IEEE, 2023, pp. 1--9

    2023

  16. [24]

    Yan et al., ``On the viability of using llms for sw/hw co-design: An example in designing cim dnn accelerators,'' in 2023 IEEE 36th International System-on-Chip Conference (SOCC)

    Z. Yan et al., ``On the viability of using llms for sw/hw co-design: An example in designing cim dnn accelerators,'' in 2023 IEEE 36th International System-on-Chip Conference (SOCC). 1em plus 0.5em minus 0.4em IEEE, 2023, pp. 1--6

    2023

  17. [25]

    Liang et al., ``Unleashing the potential of llms for quantum computing: A study in quantum architecture design,'' arXiv preprint arXiv:2307.08191, 2023

    Z. Liang et al., ``Unleashing the potential of llms for quantum computing: A study in quantum architecture design,'' arXiv preprint arXiv:2307.08191, 2023

    2023 arXiv

  18. [26]

    Li et al., ``Specllm: Exploring generation and review of vlsi design specification with large language model,'' arXiv preprint arXiv:2401.13266, 2024

    M. Li et al., ``Specllm: Exploring generation and review of vlsi design specification with large language model,'' arXiv preprint arXiv:2401.13266, 2024

    2024 arXiv

  19. [27]

    Xu et al., ``Sns's not a synthesizer: a deep-learning-based synthesis predictor,'' in Proceedings of the 49th Annual International Symposium on Computer Architecture, 2022, pp

    C. Xu et al., ``Sns's not a synthesizer: a deep-learning-based synthesis predictor,'' in Proceedings of the 49th Annual International Symposium on Computer Architecture, 2022, pp. 847--859

    2022

  20. [28]

    Zhang et al., ``Grannite: Graph neural network inference for transferable power estimation,'' in 2020 57th ACM/IEEE Design Automation Conference (DAC)

    Y. Zhang et al., ``Grannite: Graph neural network inference for transferable power estimation,'' in 2020 57th ACM/IEEE Design Automation Conference (DAC). 1em plus 0.5em minus 0.4em IEEE, 2020, pp. 1--6

    2020

  21. [29]

    M. Rakesh et al., ``Graspe: Accurate post-synthesis power estimation from rtl using graph representation learning,'' in 2023 IEEE International Symposium on Circuits and Systems (ISCAS). 1em plus 0.5em minus 0.4em IEEE, 2023, pp. 1--5

    2023

  22. [30]

    Khan et al., ``Deepseq: Deep sequential circuit learning,'' in 2024 Design, Automation & Test in Europe Conference & Exhibition (DATE)

    S. Khan et al., ``Deepseq: Deep sequential circuit learning,'' in 2024 Design, Automation & Test in Europe Conference & Exhibition (DATE). 1em plus 0.5em minus 0.4em IEEE, 2024, pp. 1--2

    2024

  23. [31]

    B. L. Miller, D. E. Goldberg et al., ``Genetic algorithms, tournament selection, and the effects of noise,'' Complex systems, vol. 9, no. 3, pp. 193--212, 1995

    1995

  24. [32]

    L. Feng et al., ``Gandse: Generative adversarial network-based design space exploration for neural network accelerator design,'' ACM Transactions on Design Automation of Electronic Systems, vol. 28, no. 3, pp. 1--20, 2023

    2023

  25. [33]

    Zhang, P

    C. Zhang, P. Li, G. Sun, Y. Guan, B. Xiao, and J. Cong, ``Dnbuilder: Automatic mapping of diverse cnn models on fpga with memory and computation optimization,'' in Proceedings of the FPGA, 2018, pp. 117--126

    2018

  26. [34]

    G. Zhou, J. Zhou, and H. Lin, ``Research on nvidia deep learning accelerator,'' in 2018 12th IEEE International Conference on Anti-counterfeiting, Security, and Identification (ASID). 1em plus 0.5em minus 0.4em IEEE, 2018, pp. 192--195

    2018

  27. [35]

    Rombach et al., ``High-resolution image synthesis with latent diffusion models,'' 2022

    R. Rombach et al., ``High-resolution image synthesis with latent diffusion models,'' 2022. [Online]. Available: https://arxiv.org/abs/2112.10752

    2022 arXiv

  28. [36]

    Kong et al., ``Diffwave: A versatile diffusion model for audio synthesis,'' in International Conference on Learning Representations, 2021

    Z. Kong et al., ``Diffwave: A versatile diffusion model for audio synthesis,'' in International Conference on Learning Representations, 2021

    2021

  29. [37]

    Ho et al., ``Denoising diffusion probabilistic models,'' in Advances in Neural Information Processing Systems, 2020

    J. Ho et al., ``Denoising diffusion probabilistic models,'' in Advances in Neural Information Processing Systems, 2020

    2020

  30. [38]

    Dhariwal et al., ``Diffusion models beat gans on image synthesis,'' Advances in Neural Information Processing Systems, vol

    P. Dhariwal et al., ``Diffusion models beat gans on image synthesis,'' Advances in Neural Information Processing Systems, vol. 34, 2021

    2021

  31. [39]

    Radford, J

    A. Radford, J. Wu, R. Child, D. Luan, D. Amodei, and I. Sutskever, ``Language models are unsupervised multitask learners,'' OpenAI Blog, vol. 1, no. 8, 2019

    2019

  32. [40]

    Du et al., ``Shidiannao: Shifting vision processing closer to the sensor,'' in Proceedings of the 42nd annual international symposium on computer architecture, 2015, pp

    Z. Du et al., ``Shidiannao: Shifting vision processing closer to the sensor,'' in Proceedings of the 42nd annual international symposium on computer architecture, 2015, pp. 92--104

    2015

  33. [41]

    Balasubramonian, A

    R. Balasubramonian, A. B. Kahng, N. Muralimanohar, A. Shafiee, and V. Srinivas, ``Cacti 7: New tools for interconnect exploration in innovative off-chip memories,'' ACM Transactions on Architecture and Code Optimization (TACO), vol. 14, no. 2, pp. 1--25, 2017

    2017

  34. [42]

    Peng et al., ``Dnn+ neurosim v2

    X. Peng et al., ``Dnn+ neurosim v2. 0: An end-to-end benchmarking framework for compute-in-memory accelerators for on-chip training,'' IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, vol. 40, no. 11, pp. 2306--2319, 2020

    2020

  35. [43]

    Loshchilov and F

    I. Loshchilov and F. Hutter, ``Decoupled weight decay regularization,'' in International Conference on Learning Representations (ICLR), 2019. [Online]. Available: https://openreview.net/forum?id=Bkg6RiCqY7

    2019

  36. [44]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, ``Imagenet: A large-scale hierarchical image database,'' in 2009 IEEE Conference on Computer Vision and Pattern Recognition. 1em plus 0.5em minus 0.4em IEEE, 2009, pp. 248--255

    2009

  37. [45]

    T. Krishna et al., ``An open-source silicon testbed for the rapid evaluation of chiplet-based systems,'' in Proceedings of the IEEE International Symposium on High-Performance Computer Architecture (HPCA). 1em plus 0.5em minus 0.4em IEEE, 2018

    2018

  38. [46]

    Heo and H

    Y. Heo and H. Kim, ``A survey on chiplet-based computing systems,'' ACM Computing Surveys (CSUR), vol. 55, no. 12, pp. 1--35, 2023

    2023

  39. [47]

    Jain et al., ``Multi-chiplet systems for accelerating transformer-based models,'' in Proceedings of the 56th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO)

    A. Jain et al., ``Multi-chiplet systems for accelerating transformer-based models,'' in Proceedings of the 56th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO). 1em plus 0.5em minus 0.4em IEEE, 2023

    2023

  40. [48]

    Z. Wang, P. S. Nalla, J. Sun, A. A. Goksoy, S. K. Mandal, J.-s. Seo, V. A. Chhabria, J. Zhang, C. Chakrabarti, U. Y. Ogras et al., ``Hisim: Analytical performance modeling and design space exploration of 2.5 d/3d integration for ai computing,'' IEEE Transactions on Computer-Ai...

    2025

  41. [49]

    Touvron, T

    H. Touvron, T. Lavril, G. Izacard, T. Lacroix, B. Rozière, N. Goyal, E. Hambro, F. Azhar, A. Rodriguez, A. Joulin, E. Grave, and G. Lample, ``Llama 2: Open foundation and fine-tuned chat models,'' 2023

    2023

  42. [50]

    Zhang, S

    S. Zhang, S. Roller, N. Goyal, M. Artetxe, M. Chen, S. Chen, C. Dewan, M. Diab, X. Li, X. V. Lin, T. Mihaylov, M. Ott, S. Shleifer, K. Shuster, D. Simig, P. S. Koura, A. Sridhar, T. Wang, and L. Zettlemoyer, ``Opt: Open pre-trained transformer language models,'' 2022

    2022

  43. [51]

    Devlin, M.-W

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, ``Bert: Pre-training of deep bidirectional transformers for language understanding,'' in Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics (NAACL). 1em plus 0.5em ...

    2019 arXiv

  44. [52]

    Xilinx , UltraScale Architecture and Product Data Sheet: Overview (DS890), AMD/Xilinx, 2022, 16 nm FinFET+, scalable architecture including Virtex UltraScale+ family with up to 12,288 DSP slices and 3.78M logic cells

    2022

  45. [53]

    Rep., 2025, logic cells, DSP slice count, block RAM/UltraRAM capacity, and transceiver I/O stats for VU13P

    ------, ``Virtex ultrascale+ fpga product brief,'' AMD/Xilinx, Tech. Rep., 2025, logic cells, DSP slice count, block RAM/UltraRAM capacity, and transceiver I/O stats for VU13P

    2025

  46. [54]

    AMD/Xilinx , ``Proven power reduction with amd ultrascale+ fpgas,'' 2025, ultraScale+ architecture innovations in process, power gating, and performance-per-watt improvements

    2025

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.