Metraipolis › archive › dc_366d252257d2

The Canonical Parallel Form as a Substrate for Parallelizing Compilers and Agentic Optimizers — arXiv:2609.28724

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.28724
License (as archived): CC0-1.0
Archived: 2026-10-08T04:51:13.371Z
SHA-256 of the archived text: 2f809906f859b6d14650eceb90d1ad1f6874026a1360bc5fe639d38b6208aeae
Stable URL of this page: https://metraipolis.com/documento/dc_366d252257d2

How to cite

The Canonical Parallel Form as a Substrate for Parallelizing Compilers and Agentic Optimizers — arXiv:2609.28724 — arXiv (en), https://arxiv.org/abs/2609.28724 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_366d252257d2 (Metraipolis archive dc_366d252257d2, sha256 2f809906f859b6d14650eceb90d1ad1f6874026a1360bc5fe639d38b6208aeae)

BibTeX

@misc{dc366d252257d2,
  title = {The Canonical Parallel Form as a Substrate for Parallelizing Compilers and Agentic Optimizers — arXiv:2609.28724},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.28724},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_366d252257d2; sha256 2f809906f859b6d14650eceb90d1ad1f6874026a1360bc5fe639d38b6208aeae},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.28724. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Imperative code fixes an execution order the computation does not require, and a parallelizing compiler must prove which parts of that order it can remove. We introduce the Canonical Parallel Form (CPF), a device-neutral program state from which every ordering constraint our analyses prove unnecessary has been removed. CPF is reached by output-preserving normalization, by lifting semantic operations such as tensor contractions, and by deriving parallelism in three levels ordered by decidability: syntactic subscript tests, exact affine dependence tests over integer sets, and SMT queries over non-linear integer arithmetic, which also admit parallelism guarded behind a runtime check. Heuristics can then specialize the canonical form for each architecture. Across 248 loop-level reasoning kernels on an AMD MI300A, CPF reaches 4.4x over Numba on its 24 Zen 4 cores and 25.4x on its CDNA 3 GPU. Against the other auto-parallelizing optimizers, CPF is 2.9x faster on the CPU and 8.7x on the GPU than DaCe's own auto-parallelizer, 1.9x faster than Pluto, and 1.5x faster than PPCG on the affine subset of the kernels. Because the pipeline is deterministic, CPF also serves as an agent's starting source, cutting the token cost per kernel by up to a factor of 2.72x while leaving the achieved speed-up unchanged, since the coding agents reason less about parallelism.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-25 in cs.PF).
Categorie arXiv: Performance (cs.PF).
Pagina del lavoro: https://arxiv.org/abs/2609.28724 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T04:28:54.414Z dalla pagina https://arxiv.org/catchup/cs.PF/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).