Pith. sign in

REVIEW 3 major objections 3 minor 35 references

Revisit Choice Network for Synthesis and Technology Mapping

T0 review · 3 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Cristal constructs Boolean choice networks with fewer but higher-quality choices, using equality saturation and priority-ranking selection to improve post-mapping area and delay over the state-of-the-art ABC implementation.

desk verdict Equality-saturation-based choice network construction is a plausible and timely idea, but the posted full text is unreadable mojibake, so current claims are unverifiable. read the letter →

arxiv 2508.14068 v1 pith:VAWFLC3S submitted 2025-08-04 cs.AR cs.AI

Chen Chen , Jiaqi Yin , Cunxi Yu This is my paper · ORCID
classification cs.ARcs.AI
keywords choicenetworktechnologymappinglogicsynthesisequalitysaturationstructuralmutationpriorityrankingABCBooleanoptimization
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 the value of a Boolean choice network lies in the quality of its choices, not their number, and that quality can be engineered directly. It introduces Cristal, a construction flow that searches representative logic cones, mutates them with equality saturation to generate diverse equivalent structures, and selects a compact set of high-ranked choices. The paper claims that after technology mapping, Cristal's networks outperform ABC's choice network construction, with average area/delay improvements of 3.85%/8.35% in delay-oriented mode and 0.11%/2.74% in area-oriented mode, plus a 63.77% runtime reduction on large circuits. A sympathetic reader would care because this suggests existing choice-network flows leave substantial mapping quality on the table by ignoring choice quality.

What carries the argument

The load-bearing object is the combination of equality saturation and priority-ranking selection. Equality saturation grows a set of equivalent rewrites into a large pool of candidate logic structures; the priority-ranking step then scores each candidate and keeps a small number of high-value choices, so the final choice network has fewer decision points that are more likely to help the mapper. This replaces the older snapshot-then-merge flow, where choices are harvested from independent optimization runs and used without ranking.

What would settle it

Run Cristal on a held-out set of large combinational circuits with a deliberately small equality-saturation budget (fewer iterations or a restricted rewrite set) and check whether post-mapping area/delay improvements fall toward zero; if the ranking cannot hold up with a modest pool, the pool's diversity is the real driver.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central finding is that a choice network built from fewer, deliberately selected choices can map better than one built from many unranked choices. The mechanism is a three-stage pipeline: representative logic cones are isolated, equality saturation is used to produce structurally diverse but functionally equivalent versions of those cones, and a priority-ranking step scores and selects the most promising choices before the network is constructed and validated. The paper reports that this pipeline beats the established choice-network construction in ABC on post-mapping area and delay across IWLS 2005, ISCAS'89, and EPFL benchmarks, and it reduces runtime substantially on the largest cases.

Load-bearing premise

The measured gains depend on equality saturation producing a sufficiently diverse and representative pool of candidate structures; if that pool does not contain the useful variants, the ranking step cannot compensate and the improvements will not generalize beyond the tested benchmarks.

Editorial extensions

If this is right

  • Technology mapping can start from a smaller, curated choice network and achieve better area/delay, making the synthesis loop faster.
  • The quality-focused construction transfers to other choice-network clients, such as equivalence checking and lossless synthesis.
  • Equality saturation becomes a practical generator of structural diversity in logic synthesis, not just a term-rewriting engine.
  • Ranking-based choice selection can be reused as a post-processing step for any flow that produces many candidate equivalent structures.

Reading between the lines

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

  • The paper's quality-over-quantity principle suggests a new objective for choice-network research: directly optimizing the distribution of choices rather than the pool size; one testable extension is learning the ranking metric from mapping data.
  • The 63.77% runtime reduction hints that the bottleneck in choice-network construction is not the search but the validation/ranking; if so, the same flow could be embedded iteratively inside a logic synthesis loop where the network is rebuilt at each pass.
  • Because the ranking is applied per cone, the approach should be localizable: a mapper could request choices for only critical cones, connecting choice-network construction to critical-path analysis.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. This paper claims a new choice network construction method, Cristal, that uses equality saturation to generate diverse logic structures and a priority-ranking selection to keep fewer but higher-quality choices, leading to improved area/delay after technology mapping compared with ABC. The abstract reports specific percentage improvements and a runtime reduction. However, the full text supplied for review is a garbled mojibake encoding from the first page onward, so none of the algorithmic details, equations, or tables are readable, and the claims cannot be checked.

Significance. If the reported numbers are correct and reproducible, the contribution would be practically valuable for technology mapping, since choice networks are a recognized technique. The evaluation appears to be against an external baseline (ABC) on standard benchmark suites, which is the right non-circular design. The work also offers concrete, falsifiable predictions (the specific improvement percentages). These strengths, however, are unusable until a readable version is available.

major comments (3)
  1. [Full text (entire document)] The posted full text is an unreadable mojibake encoding from the first page onward, with no recoverable equations, algorithms, tables, or experimental setup. Because the central claim rests entirely on the described methodology and measurements, this makes the manuscript impossible to assess in its current form. The authors need to provide a readable version, and the review should be re-run on that version.
  2. [Abstract] The abstract reports specific average improvements (3.85%/8.35% area/delay in delay-oriented mode, 0.11%/2.74% in area-oriented mode, and a 63.77% runtime reduction) but gives no error bars, circuit counts, or per-benchmark breakdown. Without these, the reader cannot tell whether the gains are consistent or driven by outliers; this information must appear in the experimental section of a readable manuscript.
  3. [Abstract] The claim that equality saturation yields a sufficiently diverse and representative pool of candidate structures is central to the method, but the abstract does not specify the equality saturation configuration (e.g., cost function, rewrite rules, iteration budget) or how the pool size affects the priority-ranking selection. This omission prevents evaluation of the generalization risk, even after the text is decodable.
minor comments (3)
  1. [Abstract] The abstract does not state the ABC version used for comparison or the CPU/clock configuration for runtime measurements.
  2. [Abstract] The phrase 'post-mapping stage' should be defined (e.g., after technology mapping, what statistical comparison is used).
  3. [Abstract] The '63.77% runtime reduction on large-scale cases' is not tied to a threshold for 'large-scale'; please clarify.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity demonstrable from available evidence; the posted full text is garbled, and the abstract's comparison against ABC is externally grounded.

full rationale

The only readable portion of the posted artifact is the abstract, which reports post-mapping area/delay and runtime comparisons of Cristal against ABC's choice network construction on IWLS 2005, ISCAS'89, and EPFL benchmarks. Those are external baselines and standard benchmark suites, so the central claim is not self-referential by construction: the improvements are measured against an independent tool. No equation, algorithm, or fitted parameter is recoverable from the posted full text, which consists of mojibake from the first page onward. Under the hard rule that circularity may be claimed only when the specific reduction can be quoted from the paper, no such circular step can be exhibited. The absence of readable experimental detail is a verifiability problem, not a demonstrated circularity. The appropriate finding is therefore no significant circularity, with the caveat that the full derivation chain cannot currently be audited from the posted artifact.

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

Review limited to abstract; full text unavailable in readable form. No technical details on hyperparameters or assumptions are provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Revisit Choice Network for Synthesis and Technology Mapping." pith.science (2026). https://pith.science/paper/VAWFLC3S

@misc{pith2026250814068,
  author       = {Pith},
  title        = {Pith review of: Revisit Choice Network for Synthesis and Technology Mapping},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VAWFLC3S}},
  note         = {Machine review of arXiv:2508.14068}
}
read the original abstract

Choice network construction is a critical technique for alleviating structural bias issues in Boolean optimization, equivalence checking, and technology mapping. Previous works on lossless synthesis utilize independent optimization to generate multiple snapshots, and use simulation and SAT solvers to identify functionally equivalent nodes. These nodes are then merged into a subject graph with choice nodes. However, such methods often neglect the quality of these choices, raising the question of whether they truly contribute to effective technology mapping. This paper introduces Cristal, a novel methodology and framework for constructing Boolean choice networks. Specifically, Cristal introduces a new flow of choice network-based synthesis and mapping, including representative logic cone search, structural mutation for generating diverse choice structures via equality saturation, and priority-ranking choice selection along with choice network construction and validation. Through these techniques, Cristal constructs fewer but higher-quality choices. Our experimental results demonstrate that Cristal outperforms the state-of-the-art Boolean choice network construction implemented in ABC in the post-mapping stage, achieving average reductions of 3.85%/8.35% (area/delay) in delay-oriented mode, 0.11%/2.74% in area-oriented mode, and a 63.77% runtime reduction on large-scale cases across a diverse set of combinational circuits from the IWLS 2005, ISCAS'89, and EPFL benchmark suites.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 33 canonical work pages

  1. [1]

    Amar \'u , P.-E

    L. Amar \'u , P.-E. Gaillardon, and G. De Micheli. The EPFL combinational benchmark suite. In IWLS , number CONF, 2015

    work page 2015

  2. [2]

    Amarú, P.-E

    L. Amarú, P.-E. Gaillardon, and G. De Micheli. Majority-inverter graph: A new paradigm for logic optimization. TCAD , 35(5):806--819, 2016

    work page 2016

  3. [3]

    Brglez, D

    F. Brglez, D. Bryan, and K. Kozminski. Combinational profiles of sequential benchmark circuits. In IEEE International Symposium on Circuits and Systems (ISCAS) , pages 1929--1934. IEEE, 1989

    work page 1929

  4. [4]

    Y. Cai, K. Yang, C. Deng, C. Yu, and Z. Zhang. Smoothe: Differentiable e-graph extraction. In Proceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1 , pages 1020--1034, 2025

    work page 2025

  5. [5]

    Chatterjee

    S. Chatterjee. On algorithms for technology mapping . University of California, Berkeley, 2007

    work page 2007

  6. [6]

    Chatterjee, A

    S. Chatterjee, A. Mishchenko, R. K. Brayton, X. Wang, and T. Kam. Reducing structural bias in technology mapping. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems , 25(12):2894--2903, 2006

    work page 2006

  7. [7]

    C. Chen, G. HU, C. Yu, Y. Ma, and H. Zhang. E-morphic: Scalable equality saturation for structural exploration in logic synthesis, 2025

    work page 2025

  8. [8]

    C. Chen, G. Hu, D. Zuo, C. Yu, Y. Ma, and H. Zhang. E-syn: E-graph rewriting with technology-aware cost functions for logic synthesis. In Proceedings of the 61st ACM/IEEE Design Automation Conference , pages 1--6, 2024

    work page 2024

Show all 35 references
  1. [9]

    Chen and J

    G. Chen and J. Cong. Simultaneous logic decomposition with technology mapping in fpga designs. In Proceedings of the 2001 ACM/SIGDA ninth international symposium on Field programmable gate arrays , pages 48--55, 2001

    2001

  2. [10]

    Cheng, S

    J. Cheng, S. Coward, L. Chelini, R. Barbalho, and T. Drane. Seer: Super-optimization explorer for high-level synthesis using e-graph rewriting. In Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Vol...

    2024

  3. [11]

    V. A. Chhabria, W. Jiang, A. B. Kahng, R. Liang, H. Ren, S. S. Sapatnekar, and B.-Y. Wu. Openroad and circuitops: Infrastructure for ml eda research and education. In 2024 IEEE 42nd VLSI Test Symposium (VTS) , pages 1--4, 2024

    2024

  4. [12]

    L. T. Clark, V. Vashishtha, L. Shifren, A. Gujja, S. Sinha, B. Cline, C. Ramamurthy, and G. Yeric. Asap7: A 7-nm finfet predictive process design kit. Microelectronics Journal , 53:105--115, 2016

    2016

  5. [13]

    Cohen, Y

    I. Cohen, Y. Huang, J. Chen, J. Benesty, J. Benesty, J. Chen, Y. Huang, and I. Cohen. Pearson correlation coefficient. Noise reduction in speech processing , pages 1--4, 2009

    2009

  6. [14]

    J. Cong, C. Wu, and Y. Ding. Cut ranking and pruning: Enabling a general and efficient fpga mapping solution. In Proceedings of the 1999 ACM/SIGDA seventh international symposium on Field programmable gate arrays , pages 29--35, 1999

    1999

  7. [15]

    Coward, G

    S. Coward, G. A. Constantinides, and T. Drane. Automatic datapath optimization using e-graphs. In 2022 IEEE 29th Symposium on Computer Arithmetic (ARITH) , pages 43--50. IEEE, 2022

    2022

  8. [16]

    Coward, G

    S. Coward, G. A. Constantinides, and T. Drane. Automating constraint-aware datapath optimization using e-graphs. In DAC , 2023

    2023

  9. [17]

    Coward, E

    S. Coward, E. Morini, B. Tan, T. Drane, and G. A. Constantinides. Datapath verification via word-level e-graph rewriting. In 2023 Formal Methods in Computer-Aided Design (FMCAD) , pages 92--100. IEEE, 2023

    2023

  10. [18]

    X. Gao, B. Xiao, D. Tao, and X. Li. A survey of graph edit distance. Pattern Analysis and applications , 13:113--129, 2010

    2010

  11. [19]

    E. I. Goldberg, M. R. Prasad, and R. K. Brayton. Using sat for combinational equivalence checking. In Proceedings Design, Automation and Test in Europe. Conference and Exhibition 2001 , pages 114--121. IEEE, 2001

    2001

  12. [20]

    Grosnit, M

    A. Grosnit, M. Zimmer, R. Tutunov, X. Li, L. Chen, F. Yang, M. Yuan, and H. Bou-Ammar. Lightweight structural choices operator for technology mapping. In 2023 60th ACM/IEEE Design Automation Conference (DAC) , pages 1--6. IEEE, 2023

    2023

  13. [21]

    Haaswijk, M

    W. Haaswijk, M. Soeken, L. Amarù, P.-E. Gaillardon, and G. De Micheli. A novel basis for logic rewriting. In ASP-DAC , pages 151--156, 2017

    2017

  14. [22]

    Z. Hu, H. Pan, Y. Xia, L. Wang, and Z. Chu. Mixed structural choice operator: Enhancing technology mapping with heterogeneous representations, 2025

    2025

  15. [23]

    Lehman, Y

    E. Lehman, Y. Watanabe, J. Grodstein, and H. Harkness. Logic decomposition during technology mapping. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems , 16(8):813--834, 1997

    1997

  16. [24]

    Mailhot and G

    F. Mailhot and G. De Micheli. Technology mapping using boolean matching. In Proc. Eur. Conf. Design Automation , pages 180--185, 1990

    1990

  17. [25]

    Mishchenko and R

    A. Mishchenko and R. Brayton. Recording synthesis history for sequential verification. In 2008 Formal Methods in Computer-Aided Design , pages 1--8. IEEE, 2008

    2008

  18. [26]

    Mishchenko, R

    A. Mishchenko, R. Brayton, S. Jang, and V. Kravets. Delay optimization using sop balancing. In 2011 IEEE/ACM International Conference on Computer-Aided Design (ICCAD) , pages 375--382. IEEE, 2011

    2011

  19. [27]

    Mishchenko, S

    A. Mishchenko, S. Chatterjee, and R. Brayton. Dag-aware aig rewriting a fresh look at combinational logic synthesis. In Proceedings of the 43rd annual Design Automation Conference , pages 532--535, 2006

    2006

  20. [28]

    Mishchenko, S

    A. Mishchenko, S. Chatterjee, and R. Brayton. Improvements to technology mapping for lut-based fpgas. In Proceedings of the 2006 ACM/SIGDA 14th international symposium on Field programmable gate arrays , pages 41--49, 2006

    2006

  21. [29]

    Ustun, I

    E. Ustun, I. San, J. Yin, C. Yu, and Z. Zhang. Impress: Large integer multiplication expression rewriting for fpga hls. In 2022 IEEE 30th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM) , pages 1--10. IEEE, 2022

    2022

  22. [30]

    Willsey, C

    M. Willsey, C. Nandi, Y. R. Wang, O. Flatt, Z. Tatlock, and P. Panchekha. Egg: Fast and extensible equality saturation. In POPL , volume 5, pages 1--29, New York, NY, USA, 2021. ACM

    2021

  23. [31]

    C. Wolf. Yosys open synthesis suite. https://yosyshq.net/yosys/

  24. [32]

    J. Yin, Z. Song, C. Chen, Q. Hu, and C. Yu. Boole: Exact symbolic reasoning via boolean equality saturation. arXiv preprint arXiv:2504.05577 , 2025

    2025 arXiv

  25. [33]

    , " * write output.state after.block = add.period write

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION in...

  26. [34]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  27. [35]

    https://yosyshq.net/yosys/

    11em plus .33em minus .07em 4000 4000 100 4000 4000 500 `\.=1000 = #1 \@IEEEnotcompsoconly \@IEEEcompsoconly #1 * [1] 0pt [0pt][0pt] #1 * [1] 0pt [0pt][0pt] #1 * \| ** #1 \@IEEEauthorblockNstyle \@IEEEcompsocnotconfonly \@IEEEauthorblockAstyle \@IEEEcompsocnotconfonly \@IEEEco...

    2006 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.