EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

parallALT: Fully Spectral GPU Accelerated Spherical Harmonic Transform Library

Dmitrii Tolmachev

ACM Transactions on Parallel Computing · 30.06.2026 · 13 (2) · 1-31

DOI: 10.1145/3807446 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACM Transactions on Parallel Computing, online erschienen am 22.06.2026. Verfasst von Dmitrii Tolmachev. Die Arbeit nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: ETH Research Commission.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This article introduces parallALT—a fully spectral spherical harmonic transform library. Developed for modern massively parallel accelerators, such as GPUs, parallALT combines close to machine-precision accuracy with exceptional performance. This article demonstrates that our fully spectral algorithm has \(O(\sqrt {L})\) (with L being the spherical harmonic degree) round trip accuracy scaling compared to O(L) scaling of the conventional quadrature approach. We also show that parallALT implementation for high L ∈ [512, 16384] is up to three times faster on an Nvidia H100 GPU than the most efficient open source quadrature implementation, SHTns. Developed with the help of the in-house cross-platform runtime code generation platform, PfSolve and VkFFT libraries, parallALT will be released as an open source library in the future.

Förderung

  • ETH Research Commission

Zitieren

APA

Tolmachev, D. (2026). parallALT: Fully Spectral GPU Accelerated Spherical Harmonic Transform Library. ACM Transactions on Parallel Computing, 13(2), 1-31. https://doi.org/10.1145/3807446

Vancouver

Tolmachev D. parallALT: Fully Spectral GPU Accelerated Spherical Harmonic Transform Library. ACM Transactions on Parallel Computing. 2026;13(2):1-31. doi:10.1145/3807446

BibTeX
@article{Tolmachev2026,
  author = {Tolmachev, Dmitrii},
  title = {parallALT: Fully Spectral GPU Accelerated Spherical Harmonic Transform Library},
  journal = {ACM Transactions on Parallel Computing},
  year = {2026},
  volume = {13},
  number = {2},
  pages = {1--31},
  doi = {10.1145/3807446}
}
Kostenlos starten →