Pith. sign in

REVIEW 3 major objections 4 minor 64 references

StructMG: A Fast and Scalable Structured Algebraic Multigrid

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Structured algebraic multigrid can be cheap per iteration and reliably convergent at once, the paper argues.

desk verdict A solid structured-AMG engineering contribution with real artifacts, but the headline speedups rest on benchmark fairness that the paper itself concedes is imperfect. read the letter →

arxiv 2506.21932 v1 pith:WD5LRWYU submitted 2025-06-27 math.NA cs.CEcs.NAcs.PF

classification math.NAcs.CEcs.NAcs.PF MSC 65F1065F0865N5565Y05
keywords structuredgridalgebraicmultigridparallelpreconditionerGalerkincoarseningsparsetriangularsolveGauss-SeidelsmootherincompleteLUlarge-scalelinearsystems
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

The paper tries to establish that a structured algebraic multigrid can be both cheap per iteration and reliably convergent, instead of forcing users to choose between few iterations and fast iterations. From the classical 'multigrid seesaw' it derives three principles: keep stencil data layouts, coarsen in all dimensions to cut grid and operator complexity, and pair that with dependence-preserving smoothers. These are implemented in StructMG, which builds the coarse-grid hierarchy automatically from the fine-level matrix alone. On six 3D problems from radiation hydrodynamics, reservoir simulation, weather prediction, and elasticity, run on ARM and x86 clusters, StructMG reports the fastest time-to-solution in every case, with average speedups of 15.5x, 5.5x, 6.7x, and 7.3x over SMG, PFMG, SysPFMG, and BoomerAMG. A careful reader should take the speedup numbers as conditional on how well the baselines were tuned, but the paper's core claim is that the trade-off is not intrinsic.

What carries the argument

The load-bearing device is the stencil-based triple-matrix product for Galerkin coarsening, $A_C = R A_F P$, implemented by symbolic derivation and code generation: the product is reinterpreted as chains of influence running through fine-grid elements, the formulas for every combination of stencil patterns are derived symbolically, and fused C++ kernels are generated once and reused for all matrices with those patterns. This keeps the coarse operator in stencil form, avoids CSR storage and general SpGEMM, and makes full multi-dimensional coarsening practical despite thousands of influence chains. The other device is a unified level-scheduled sparse triangular solver (SpTRSV) framework that projects 3D stencils to 2D, schedules columns in dependency levels, sparsifies synchronization by removing two-hop and intra-thread dependencies, and uses counter-based point-to-point synchronization so that Gauss-Seidel and ILU smoothers run in parallel while computing exactly the same values as their serial versions.

What would settle it

Run all six problems with the same tolerances but replace hand-picked baseline settings by an automated search of, say, a thousand parameter combinations per solver, and compare best total time; if BoomerAMG or SysPFMG then matches StructMG's total time on a majority of cases, the claimed 7.3x and 6.7x averages are artifacts of configuration. Separately, solve a 3D problem whose anisotropy axis rotates across the grid; if StructMG's ILU smoother needs more than a small constant factor more iterations as anisotropy strengthens, the convergence part of the claim is not general.

Watch

Extended reading notes

Core claim

StructMG's central claim is that the defect of earlier structured multigrids is an imbalance on the 'multigrid seesaw': SMG spends almost everything on coarse-grid correction through plane smoothing, while PFMG and SysPFMG spend little on smoothing but pay for it with weak 1D semi-coarsening, high grid complexity, and many iterations. StructMG asserts that fixed-stride coarsening in all dimensions lowers grid complexity in 3D to about $8/7$ instead of roughly $2$, cutting memory traffic, levels, and communication rounds, and that this loss in coarse-grid quality can be compensated by smoothers that preserve the original dependency order, namely point and line Gauss-Seidel and ILU variants executed through a shared parallel sparse triangular solver. The paper's evidence is end-to-end: on all six test problems, on both platforms, StructMG is fastest in total time and has the best strong and weak scaling in the reported runs. The result, if correct, is that structure and algebra are not in tension: a stencil-based algebraic hierarchy can be built automatically and still beat general-purpose and specialized solvers.

Load-bearing premise

The load-bearing premise is that the four baseline solvers were configured close to their best attainable performance; if they were not, the reported average speedups overstate StructMG's real advantage.

Editorial extensions

If this is right

  • Full 3D coarsening cuts the number of levels roughly by a factor of 3-4 relative to SMG and PFMG, so each V-cycle needs fewer communication rounds and less halo exchange.
  • Lower grid and operator complexities translate directly into less memory traffic per iteration, which is the dominant cost in stencil solvers.
  • Because the GS and ILU smoothers preserve dependency order, iteration counts in the paper's tests do not worsen as thread counts grow, unlike hybrid Jacobi/GS smoothers in unstructured AMG.
  • The code-generation approach means new stencil patterns or new interpolation choices can be added by regenerating fused kernels rather than by writing thousands of lines of hand-derived formulas.
  • The same design principles can transfer to semi-structured grids and GPUs, extending the benefit beyond purely rectangular structured grids.

Reading between the lines

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

  • If the speedups survive exhaustive autotuning of the baselines, structured solvers could replace unstructured AMG as the default preconditioner on logically rectangular grids; the reported averages of 15.5x, 5.5x, 6.7x, and 7.3x are optimistic to the extent the baselines were hand-tuned rather than fully searched.
  • The symbolic derivation strategy is general: the same idea could generate fused coarsening kernels for coarsening factors other than 2, rotated stencils, or variable-order interpolation, which the paper does not explore.
  • The paper demonstrates convergence on six cases but does not prove it for arbitrary anisotropy; a natural stress test is a 3D problem whose anisotropy direction rotates continuously through the grid, where fixed-stride coarsening and ILU(0) may need more fill-in or line smoothing.
  • For sequences of linear systems that share the same stencil, as in many time-stepping applications, setup costs amortize over solves; the paper does not quantify this additional advantage.
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 / 4 minor

Summary. The paper introduces StructMG, a structured algebraic multigrid preconditioner designed to combine low per-iteration cost with robust convergence for structured-grid problems. The main technical contributions are a design rationale based on a 'multigrid seesaw' analogy, a stencil-based symbolic derivation and code-generation method for the Galerkin triple-matrix product, and a unified parallel sparse triangular solver framework supporting dependence-preserving Gauss-Seidel and ILU smoothers. The paper evaluates StructMG against hypre's SMG, PFMG, SysPFMG, and BoomerAMG on six benchmark problems from radiation hydrodynamics, reservoir simulation, weather prediction, and solid mechanics, on ARM and X86 platforms, reporting average time-to-solution speedups of 15.5x, 5.5x, 6.7x, and 7.3x over the four baselines, respectively.

Significance. If the reported results are robust, this is a useful and significant contribution to high-performance structured AMG. The paper includes concrete, testable claims: low grid and operator complexity from multi-dimensional coarsening, generated code that matches the stated influence-chain derivation, and an artifact with archived source code and baseline drivers. The ablation study in Section 6.3 also gives a clean test of the smoother-design claim. The main uncertainty is not the algebraic derivation, which is standard, but the fairness and reproducibility of the performance comparisons, which are load-bearing for every headline speedup in the paper.

major comments (3)
  1. [Section 5.2, Table 7, Abstract] The central speedup claims depend on the baselines being fairly configured, but Section 5.2 admits that BoomerAMG was tuned only 'after trying our best' and 'may not be the real best in the entire parameter space,' while StructMG uses per-problem hand-picked coarsening, smoother, interpolation, and relaxation settings listed in Appendix Table 8. Since the average 7.3x speedup over BoomerAMG includes factors of 12.7-22.5x on weather, this asymmetry is material. The paper should provide a systematic tuning procedure for BoomerAMG, a sensitivity analysis over its main parameters, or otherwise show that the reported advantage is not an artifact of under-configured baselines.
  2. [Table 3 vs Appendix Table 8] The problem sizes are inconsistent between the main text and the appendix. Table 3 lists Laplace as 16.8M unknowns and rhd as 2.10M, while Appendix Table 8 lists Laplace as 2.10M and rhd as 31.5M. These entries appear to be swapped. This disagreement must be resolved because the scaling curves and speedup numbers in Figures 8-9 and Table 7 cannot be checked until it is clear which problem size was actually solved for each case.
  3. [Figures 8-9 and Table 7] All timing results are reported as single values with no repeated-run statistics or variance information. The minimum reported speedup over PFMG on X86 Laplace is only 1.14x, which is close enough to run-to-run system noise that it could reverse the 'fastest in all cases' claim for that configuration. The paper should report medians or multiple repetitions with error bars, or at least state the number of repeated runs used to select the reported time.
minor comments (4)
  1. [Section 4.1] The text says the three principles are 'derived' from the multigrid seesaw, but the argument is heuristic design reasoning supported by later experiments, not a mathematical derivation. I suggest rephrasing to 'motivated by' to avoid overclaiming.
  2. [References] References [13] and [14] are duplicate entries for the same paper by Falgout and Schroder; one should be removed or replaced with a distinct relevant reference.
  3. [Table 6] The row labeled 'weather (g)' is explicitly stated to be solved by Semi-StructMG rather than by StructMG. Since this row appears in the ablation table that otherwise evaluates StructMG smoothers, its inclusion should be more clearly separated or qualified so that readers do not mistake it as a StructMG result.
  4. [Section 4.3, Equations (5)-(8)] The notation for the 'range' sets in Equations (5) and (6) is terse: readers must infer that Offset(2d9) is added as a set of coordinate offsets. A brief sentence defining set-valued addition would improve readability.

Circularity Check

0 steps flagged · score 1.0 of 10

No material circularity: the design principles are heuristic, the speedups are externally benchmarked measurements, and the self-citations are not load-bearing.

full rationale

StructMG's central claims are empirical and algebraic rather than derived from fitted inputs. The 'multigrid seesaw' quotation from [43] motivates principles P1-P3, but these principles are design heuristics, not equations that predetermine the reported convergence rates or timings; no parameter in the paper is fit to the benchmark outcome and then reported as a prediction. The Galerkin coarsening derivation (Eqs. 3-9) is a standard symbolic expansion of the identity A_C = R A_F P, and the stencil-based code generation is an implementation technique, so the setup results do not reduce to the paper's own definitions by construction. The benchmark speedups against SMG, PFMG, SysPFMG, and BoomerAMG are measurements from external runs of hypre-2.25.0, with the baselines' settings documented; the admitted 'may not be the real best' caveat for BoomerAMG is a comparison-fairness risk, not circularity. Self-citations [35, 61, 62] appear for problem statistics, interface documentation, and a GPU outlook, but none carries a load-bearing theorem or uniqueness argument, and none is used to forbid alternative designs. The Appendix A discrepancy in Laplace problem size is an internal consistency issue, not a circular derivation. Overall, no step in the paper makes a predicted quantity equal to its own input by construction.

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

The central numerical comparisons depend on the usual axioms of multigrid (stencil representation, Galerkin coarsening, smoothing), plus the paper's own domain assumptions: fixed-stride coarsening remains convergent when paired with stronger smoothers; the generated RAP formulas exhaust all influence chains; the parallel SpTRSV reproduces serial results. The per-problem hand-picked settings (coarsening strategy, smoother, interpolation weight, relaxation weight, MPI/OpenMP ratio) are free configuration choices that materially affect the reported speedups. No new physical entities are introduced.

free parameters (5)
  • Per-problem coarsening strategy = 2D semi-coarsening for weather; 3D full coarsening for other test problems
    Chosen per problem in Appendix Table 8; controls grid complexity and convergence, and qualifies the multi-dimensional coarsening claim.
  • Per-problem smoother choice = PGS for Laplace and rhd; ILU(0) for rhd, rhd-3T, and solid-3D; LGS for weather and oil-4C
    Table 6 ablation shows iteration counts vary widely, for example 1200 for PGS versus 31 for LGS on oil-4C, so the smoother selection drives the headline speedups.
  • Smoothed interpolation weight = 0.7 for rhd and rhd-3T
    Appendix Table 8, Cell_3d8SIp; a hand-set scalar that influences interpolation quality and convergence.
  • Relaxation weight = 1.0 for the listed StructMG cases
    Appendix Table 8; standard for GS and ILU but a free choice per run.
  • MPI/OpenMP ratio and process partitioning = Multiple ratios tested per core count; the best result is reported
    Section 5.3; selecting the best ratio from a set favors the reported scaling efficiency and is not a fixed algorithmic input.
assumptions (5)
  • domain assumption Fixed-stride multi-dimensional coarsening with robust smoothers yields a convergent multilevel hierarchy on structured problems
    The paper relies on Principle P3 to compensate for Principle P2 in Section 4.1; no convergence theorem is given, only empirical convergence results in Section 6.3.
  • domain assumption Symbolic derivation and code generation implement the fused Galerkin triple-matrix product exactly for all supported stencil masks
    Section 4.3 derives influence chains for one example and generates code, but the generated kernels are not formally verified against a reference implementation for every mask combination.
  • domain assumption The level-based SpTRSV with sparsified and intra-thread dependence elimination produces results identical to serial GS and ILU
    Section 4.4 claims parallel results equal serial results, but no machine-checked or exhaustive verification is provided; this is essential for the convergence comparisons.
  • domain assumption Structured matrices stored in SG-DIA with zero-padded boundaries have no indirect addressing and memory volume O(N*nzpr)
    Principle P1 in Section 4.1; underpins all kernel-performance comparisons and the claim that stencil-like access is faster than CSR.
  • domain assumption The hand-tuned hypre baseline settings are representative of the best achievable performance of those solvers
    Section 5.2 admits BoomerAMG settings may not be the real best in the entire parameter space; if false, the speedup factors are inflated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of StructMG: A Fast and Scalable Structured Algebraic Multigrid." pith.science (2026). https://pith.science/paper/WD5LRWYU

@misc{pith2026250621932,
  author       = {Pith},
  title        = {Pith review of: StructMG: A Fast and Scalable Structured Algebraic Multigrid},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WD5LRWYU}},
  note         = {Machine review of arXiv:2506.21932}
}
read the original abstract

Parallel multigrid is widely used as preconditioners in solving large-scale sparse linear systems. However, the current multigrid library still needs more satisfactory performance for structured grid problems regarding speed and scalability. Based on the classical 'multigrid seesaw', we derive three necessary principles for an efficient structured multigrid, which instructs our design and implementation of StructMG, a fast and scalable algebraic multigrid that constructs hierarchical grids automatically. As a preconditioner, StructMG can achieve both low cost per iteration and good convergence when solving large-scale linear systems with iterative methods in parallel. A stencil-based triple-matrix product via symbolic derivation and code generation is proposed for multi-dimensional Galerkin coarsening to reduce grid complexity, operator complexity, and implementation effort. A unified parallel framework of sparse triangular solver is presented to achieve fast convergence and high parallel efficiency for smoothers, including dependence-preserving Gauss-Seidel and incomplete LU methods. Idealized and real-world problems from radiation hydrodynamics, petroleum reservoir simulation, numerical weather prediction, and solid mechanics, are evaluated on ARM and X86 platforms to show StructMG's effectiveness. In comparison to \textit{hypre}'s structured and general multigrid preconditioners, StructMG achieves the fastest time-to-solutions in all cases with average speedups of 15.5x, 5.5x, 6.7x, 7.3x over SMG, PFMG, SysPFMG, and BoomerAMG, respectively. StructMG also significantly improves strong and weak scaling efficiencies.

Figures

Figures reproduced from arXiv: 2506.21932 by the authors.

Figure 1
Figure 1. Strong scaling tests of hypre ’s structured AMG against unstructured AMG as preconditioners for rhd prob￾lems. Numbers of iterations are labeled near the lines. 1Vector PDE: multiple unknowns associated with each grid element. arXiv:2506.21932v1 [math.NA] 27 Jun 2025 [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. AMG overview. V-Cycle in the solve phase [61]. [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the correspondence between the [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Typical nonzero patterns of structured matrices. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Illustration of stencil-based triple-matrix product [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Illustration of SpTRSV on a structured grid. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: (a) SpMV & SpTRSV and (b) triple-matrix product [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: The breakdown of the terms in Equation (1) is shown in [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 8
Figure 8. Figure 8: Strong scalability results. Lower is better in all subfigures. Subfigures share legends at the top, y-axis labels at the left, [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Only total time and #iter are shown due to page limit. [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 5
Figure 5. Figure 5 [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 50 canonical work pages

  1. [1]

    China Meteorological Administration. 2016. GRAPES Numerical Weather Predic- tion System. Retrieved July 7, 2023 from https://www.cma.gov.cn/2011xwzx/ 2011xqxxw/2011xqxyw/202110/t20211030_4079298.html

  2. [2]

    Ashby and Robert D

    Steven F. Ashby and Robert D. Falgout. 1996. A Parallel Multigrid Preconditioned Conjugate Gradient Algorithm for Groundwater Flow Simulations. Nuclear Science and Engineering 124, 1 (1996), 145–159. doi:10.13182/NSE96-A24230

  3. [3]

    Baker, Robert D

    Allison H. Baker, Robert D. Falgout, Tzanio V. Kolev, and Ulrike Meier Yang. 2012. Scaling Hypre’s Multigrid Solvers to 100,000 Cores . Springer London, London, 261–279. doi:10.1007/978-1-4471-2437-5_13

  4. [4]

    A. H. Baker, Tz. V. Kolev, and U. M. Yang. 2010. Improving algebraic multi- grid interpolation operators for linear elasticity problems. Numerical Lin- ear Algebra with Applications 17, 2-3 (2010), 495–517. doi:10.1002/nla.688 arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/nla.688

  5. [5]

    Falgout, William Gropp, Luke N

    Amanda Bienz, Robert D. Falgout, William Gropp, Luke N. Olson, and Jacob B. Schroder. 2015. Reducing Parallel Communication in Algebraic Multigrid through Sparsification. SIAM J. Sci. Comput. 38 (2015)

  6. [6]

    Siham Boukhris, Artem Napov, and Yvan Notay. 2023. Algebraic Multi- grid Using a Stencil–CSR Hybrid Format on GPUs. SIAM Journal on Scientific Computing 45, 3 (2023), C154–C178. doi:10.1137/22M1480938 arXiv:https://doi.org/10.1137/22M1480938

  7. [7]

    BROWN, Robert D

    Peter N. BROWN, Robert D. FALGOUT, and J. E. JONES. 2000. Semicoarsening multigrid on distributed memory machines. SIAM J. Sci. Comput. 21, 5 (2000), 1823–1834. doi:10.1137/S1064827598339141

  8. [8]

    M. A. Christie and M. J. Blunt. 2001. Tenth SPE Comparative So- lution Project: A Comparison of Upscaling Techniques. SPE Reservoir Evaluation & Engineering 4, 04 (08 2001), 308–317. doi:10.2118/72469-PA arXiv:https://onepetro.org/REE/article-pdf/4/04/308/2586053/spe-72469-pa.pdf

Show all 64 references
  1. [9]

    J.E Dendy. 1982. Black box multigrid. J. Comput. Phys. 48, 3 (1982), 366–386. doi:10.1016/0021-9991(82)90057-2

  2. [10]

    Ernesto Dufrechou and Pablo Ezzatti. 2018. A New GPU Algorithm to Compute a Level Set-Based Analysis for the Parallel Solution of Sparse Triangular Systems. In 2018 IEEE International Parallel and Distributed Processing Symposium (IPDPS) . 920–929. doi:10.1109/IPDPS.2018.00101

  3. [11]

    Falgout and Jim E

    Robert D. Falgout and Jim E. Jones. 2000. Multigrid on Massively Parallel Archi- tectures. In Multigrid Methods, E. Dick (Ed.). Multigrid Methods, Vol. 5

  4. [12]

    Falgout, Ruipeng Li, Björn Sjögreen, Lu Wang, and Ulrike Meier Yang

    Robert D. Falgout, Ruipeng Li, Björn Sjögreen, Lu Wang, and Ulrike Meier Yang

  5. [14]

    Falgout and Jacob B

    Robert D. Falgout and Jacob B. Schroder. 2014. Non-Galerkin Coarse Grids for Algebraic Multigrid. SIAM J. Sci. Comput. 36, 3 (jan 2014), C309–C334. doi:10. 1137/130931539

  6. [15]

    Hormozd Gahvari, Allison Baker, Martin Schulz, Ulrike Yang, Kirk Jordan, and William Gropp. 2011. Modeling the performance of an algebraic multigrid cycle on HPC platforms. 172-181 pages

  7. [16]

    Jordan, Martin Schulz, and Ulrike Meier Yang

    Hormozd Gahvari, William Gropp, Kirk E. Jordan, Martin Schulz, and Ulrike Meier Yang. 2013. Systematic Reduction of Data Movement in Algebraic Multigrid Solvers. In Proceedings of the 2013 IEEE 27th International Symposium on Par- allel and Distributed Processing Workshops and...

  8. [17]

    Van Emden Henson and Ulrike Meier Yang. 2002. BoomerAMG: A parallel algebraic multigrid solver and preconditioner. Applied Numerical Mathematics 41 (2002), 155–177

  9. [18]

    J. E. Dendy Jr and J. D. Moulton. 2010. Black Box Multigrid with coarsening by a factor of three. Numerical Linear Algebra with Applications 17, 2-3 (2010), 577–598. doi:10.1002/nla.705 arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/nla.705

  10. [19]

    Khalil and P

    M. Khalil and P. Wesseling. 1992. Vertex-centered and cell-centered multigrid for interface problems. J. Comput. Phys. 98, 1 (1992), 1–10. doi:10.1016/0021- 9991(92)90168-X

  11. [20]

    Do Y. Kwak. 1999. V-Cycle Multigrid for Cell-Centered Finite Differences. SIAM J. Sci. Comput. 21, 2 (jan 1999), 552–564. doi:10.1137/S1064827597327310

  12. [21]

    Lawrence Livermore National Lab. 2023. BoomerAMG. Retrieved March 3, 2023 from https://hypre.readthedocs.io/en/latest/solvers-boomeramg.html

  13. [22]

    Lawrence Livermore National Lab. 2023. Documentation for hypre. Retrieved March 3, 2023 from https://hypre.readthedocs.io/en/latest

  14. [23]

    Lawrence Livermore National Lab. 2023. SplitSolve. Retrieved March 3, 2023 from https://hypre.readthedocs.io/en/latest/solvers-split.html

  15. [24]

    Lawrence Livermore National Lab. 2023. Structured multigrid in HYPRE. Re- trieved March 3, 2023 from https://hypre.readthedocs.io/en/latest/solvers-smg- pfmg.html

  16. [25]

    Kun Li, Liang Yuan, Yunquan Zhang, Yue Yue, and Hang Cao. 2022. An Efficient Vectorization Scheme for Stencil Computation. In 2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS) . 650–660. doi:10.1109/IPDPS53621. 2022.00069

  17. [26]

    Ruipeng Li and Ulrike Meier Yang. 2021. Performance Evaluation of hypre Solvers. (Feb 2021). doi:10.2172/1764323

  18. [27]

    Sadayappan, and Jason Sarich

    Daniel Lowell, Jeswin Godwin, Justin Holewinski, Deepan Karthik, Chekuri Choudary, Azamat Mametjanov, Boyana Norris, Gerald Sabin, P. Sadayappan, and Jason Sarich. 2013. Stencil-Aware GPU Optimization of Iterative Solvers. SIAM Journal on Scientific Computing 35, 5 (2013), S20...

  19. [28]

    Victor A. P. Magri, Robert D. Falgout, and Ulrike M. Yang. 2023. A New Semistructured Algebraic Multigrid Method. SIAM Journal on Scientific Computing 45, 3 (2023), S439–S460. doi:10.1137/21M1434118 arXiv:https://doi.org/10.1137/21M1434118 , , Yi Zong, Peinan Yu, Haopeng Huang...

  20. [29]

    Michael Mascagni, Amanda Bienz, William D Gropp, and Luke N Olson. 2020. Reducing Communication in Algebraic Multigrid with Multi-Step Node Aware Communication. Int. J. High Perform. Comput. Appl. 34, 5 (sep 2020), 547–561. doi:10.1177/1094342020925535

  21. [30]

    Christopher Maynard, Thomas Melvin, and Eike Hermann Müller. 2020. Multi- grid preconditioners for the mixed finite element dynamical core of the LFRic atmospheric model. Quarterly Journal of the Royal Meteorological Society 146, 733 (Oct. 2020), 3917–3936. doi:10.1002/qj.3880...

  22. [31]

    Marcus Mohr and Roman Wienands. 2004. Cell-centred multigrid revisited. Computing and Visualization in Science 7, 3-4 (2004), 129–140. doi:10.1007/s00791- 004-0137-0 identifier: 137

  23. [32]

    Maxim Naumov. 2011. Parallel Solution of Sparse Triangular Linear Systems in the Preconditioned Iterative Methods on the GPU . Technical Report Technical Report NVR-2011-001. NVIDIA Corp.„ Westford, MA, USA

  24. [33]

    Anthony Nguyen, Nadathur Satish, Jatin Chhugani, Changkyu Kim, and Pradeep Dubey. 2010. 3.5-D Blocking Optimization for Stencil Computations on Modern CPUs and GPUs. In Proceedings of the 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storag...

  25. [34]

    Society of Petroleum Engineers. 2023. SPE Comparative Solution Project. Re- trieved March 3, 2023 from https://www.spe.org/web/csp/datasets/set02.htm

  26. [35]

    Anonymous (omitted due to double-anonymous review). [n. d.]. Double- anonymous review requires us to anonymize citations to reference our unpub- lished work. ([n. d.])

  27. [36]

    Jongsoo Park, Mikhail Smelyanskiy, Narayanan Sundaram, and Pradeep Dubey

  28. [37]

    Jongsoo Park, Mikhail Smelyanskiy, Ulrike Meier Yang, Dheevatsa Mudigere, and Pradeep Dubey. 2015. High-Performance Algebraic Multigrid Solver Optimized for Multi-Core Based Distributed Parallel Systems. In Proceedings of the Inter- national Conference for High Performance Com...

  29. [38]

    China Meteorological News Press. 2014. An Introduction of GRAPES. Retrieved July 7, 2023 from https://www.cma.gov.cn/en/NewsReleases/MetInstruments/ 201403/t20140327_241784.html

  30. [39]

    Olson, and J

    Andrew Reisner, Luke N. Olson, and J. David Moulton. 2018. Scaling Struc- tured Multigrid to 500K+ Cores Through Coarse-Grid Redistribution. SIAM Journal on Scientific Computing 40, 4 (2018), C581–C604. doi:10.1137/17M1146440 arXiv:https://doi.org/10.1137/17M1146440

  31. [40]

    2003.Iterative Methods for Sparse Linear Systems (second ed.)

    Yousef Saad. 2003.Iterative Methods for Sparse Linear Systems (second ed.). Society for Industrial and Applied Mathematics. doi:10.1137/1.9780898718003

  32. [41]

    Martin H. Sadd. 2005. Elasticity: Theory, Applications, and Numerics. Academic Press. doi:10.1016/B978-0-12-605811-6.X5000-3

  33. [42]

    Steve Schaffer. 1998. A semi-coarsening multigrid method for elliptic partial differential equations with highly discontinuous and anisotropic coefficients. SIAM J. Sci. Comput. 20, 1 (1998), 228–242

  34. [43]

    K. Stuben. 2000. Algebraic Multigrid (AMG) : An Introduction With Applications

  35. [44]

    L.H. Thomas. 1949. Elliptic Problems in Linear Differential Equations over a Network. Technical Report. Watson Sci. Comput. Lab Report, Columbia University, New York, USA

  36. [45]

    Eran Treister and Irad Yavneh. 2015. Non-Galerkin Multigrid Based on Sparsified Smoothed Aggregation. SIAM J. Sci. Comput. 37 (2015)

  37. [46]

    Oosterlee, and Anton Schuller

    Ulrich Trottenberg, Cornelius W. Oosterlee, and Anton Schuller. 2001. Multigrid. Academic Press, San Diego, California, USA

  38. [47]

    Knut uhh, Florian Lemarié, Laurent inria, and Hans Burchard. 2018. The numerics of hydrostatic structured-grid coastal ocean models: State of the art and future perspectives. Ocean Modelling 125 (01 2018). doi:10.1016/j.ocemod.2018.01.007

  39. [48]

    Xinliang Wang, Weifeng Liu, Wei Xue, and Li Wu. 2018. SwSpTRSV: A Fast Sparse Triangular Solve with Sparse Level Tile Layout on Sunway Architectures. In Proceedings of the 23rd ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming (Vienna, Austria) (PPoPP ’1...

  40. [49]

    Xinliang Wang, Ping Xu, Wei Xue, Yulong Ao, Chao Yang, Haohuan Fu, Lin Gan, Guangwen Yang, and Weimin Zheng. 2018. A Fast Sparse Triangular Solver for Structured-Grid Problems on Sunway Many-Core Processor SW26010. In Proceedings of the 47th International Conference on Paralle...

  41. [50]

    Wesseling

    P. Wesseling. 1988. Cell-Centered Multigrid for Interface Problems. J. Comput. Phys. 79, 1 (1988), 85–91

  42. [51]

    Xu Xiaowen. 2019. Parallel Algebraic Multigrid Methods: State-Of-The-Art and Challenges for Extreme-Scale Applications. Journal on Numerica Methods and Computer Applications 40, 4, Article 243 (2019), 243-260 pages

  43. [52]

    Chenhao Xie, Jieyang Chen, Jesun Firoz, Jiajia Li, Shuaiwen Leon Song, Kevin Barker, Mark Raugas, and Ang Li. 2021. Fast and Scalable Sparse Triangular Solver for Multi-GPU Based HPC Architectures. In Proceedings of the 50th International Conference on Parallel Processing (Lem...

  44. [53]

    Xiaowen Xu and Zeyao Mo. 2017. Algebraic interface-based coars- ening AMG preconditioner for multi-scale sparse matrices with appli- cations to radiation hydrodynamics computation. Numerical Linear Algebra with Applications 24, 2 (2017), e2078. doi:10.1002/nla.2078 arXiv:https...

  45. [54]

    2020.𝛼Setup- AMG: an adaptive-setup-based parallel AMG solver for sequence of sparse linear systems

    Xiaowen Xu, Zeyao Mo, Xiaoqiang Yue, Hengbin An, and Shi Shu. 2020.𝛼Setup- AMG: an adaptive-setup-based parallel AMG solver for sequence of sparse linear systems. CCF Transactions on High Performance Computing 2 (2020), 98–110. doi:10.1007/s42514-020-00033-w

  46. [55]

    Ulrike Meier Yang. 2010. On long-range interpolation operators for aggressive coarsening. Numerical Linear Algebra with Applications 17, 2-3 (2010), 453–472. doi:10.1002/nla.689 arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/nla.689

  47. [56]

    Xiaojian Yang, Shengguo Li, Fan Yuan, Dezun Dong, Chun Huang, and Zheng Wang. 2023. Optimizing Multi-Grid Computation and Parallelization on Multi- Cores. In Proceedings of the 37th International Conference on Supercomputing (Orlando, FL, USA) (ICS ’23). Association for Comput...

  48. [57]

    Chensong Zhang, Shizhe Li, Li Zhao, Xiaoxing Cheng, and Yang Liu

  49. [58]

    Chensong Zhang, Shizhe Li, Li Zhao, Xiaoxing Cheng, and Yang Liu. 2023. Open- CAEPoro. Retrieved March 3, 2023 from https://github.com/OpenCAEPlus/ OpenCAEPoro/tree/main/examples/spe10

  50. [59]

    Renhe Zhang and Xueshun Shen. 2008. On the development of the GRAPES—A new generation of the national operational NWP system in China.Chinese Science Bulletin 53 (11 2008), 3429–3432. doi:10.1007/s11434-008-0462-7

  51. [60]

    Qianchao Zhu, Hao Luo, Chao Yang, Mingshuo Ding, Wanwang Yin, and Xinhui Yuan. 2021. Enabling and Scaling the HPCG Benchmark on the Newest Generation Sunway Supercomputer with 42 Million Heterogeneous Cores. In Proceedings of the International Conference for High Performance C...

  52. [61]

    Yi Zong, Peinan Yu, Haopeng Huang, and Wei Xue. 2024. FP16 Acceleration in Structured Multigrid Preconditioner for Real-World Applications. In Proceedings of the 53rd International Conference on Parallel Processing (Gotland, Sweden) (ICPP ’24). Association for Computing Machin...

  53. [62]

    Yi Zong, Chensong Zhang, Longjiang Mu, Jianchun Wang, Jian Sun, Xiaowen Xu, Xinliang Wang, Peinan Yu, and Wei Xue. 2025. Semi-StructMG: A Fast and Scalable Semi-Structured Algebraic Multigrid. In Proceedings of the 30th ACM SIGPLAN Annual Symposium on Principles and Practice o...

  54. [2014]

    In Proceedings of the 29th International Conference on Super- computing - Volume 8488 (Leipzig, Germany) (ISC 2014)

    Sparsifying Synchronization for High-Performance Shared-Memory Sparse Triangular Solver. In Proceedings of the 29th International Conference on Super- computing - Volume 8488 (Leipzig, Germany) (ISC 2014). Springer-Verlag, Berlin, Heidelberg, 124–140. doi:10.1007/978-3-319-07518-1_8

  55. [2021]

    Parallel Comput

    Porting hypre to heterogeneous computer architectures: Strategies and experiences. Parallel Comput. 108 (2021), 102840. doi:10.1016/j.parco.2021.102840

  56. [2023]

    Retrieved March 3, 2023 from https://github.com/OpenCAEPlus/OpenCAEPoro/blob/main/ examples/spe1a/csr.fasp

    Oil-application-specific tuning of algebraic multigrid. Retrieved March 3, 2023 from https://github.com/OpenCAEPlus/OpenCAEPoro/blob/main/ examples/spe1a/csr.fasp

Pith tools

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