Metraipolis › archive › dc_00f0044034cc

PENDA: An Efficient Processing Element via Norm-of-Difference for Deep Learning Accelerators — arXiv:2609.08424

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08424
License (as archived)
CC0-1.0
Archived
2026-10-16T13:14:41.019Z
SHA-256 of the archived text
74c3a98c8636d384a7746f40b38ecbb6d77b5a0d2462df1a16d6516c9fd699f1
Stable URL of this page
https://metraipolis.com/documento/dc_00f0044034cc

How to cite

PENDA: An Efficient Processing Element via Norm-of-Difference for Deep Learning Accelerators — arXiv:2609.08424 — arXiv (en), https://arxiv.org/abs/2609.08424 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_00f0044034cc (Metraipolis archive dc_00f0044034cc, sha256 74c3a98c8636d384a7746f40b38ecbb6d77b5a0d2462df1a16d6516c9fd699f1)
BibTeX
@misc{dc00f0044034cc,
  title = {PENDA: An Efficient Processing Element via Norm-of-Difference for Deep Learning Accelerators — arXiv:2609.08424},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08424},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_00f0044034cc; sha256 74c3a98c8636d384a7746f40b38ecbb6d77b5a0d2462df1a16d6516c9fd699f1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08424. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Inner product computation dominates the computational cost of deep learning models; thus, accelerating this primitive is key to improving hardware efficiency. However, most existing techniques rely on approximations, which can degrade model accuracy. To preserve exactness while optimizing hardware, this paper presents PENDA (processing element via norm-of-difference architecture), which leverages the law of cosines to recast multiplications as squared-difference operations. Replacing multiply-accumulate units with the proposed norm-of-difference units yields 11~36%, 5~48%, and 11~19% reductions in area, energy, and clock period, respectively, for the PE array of a deep learning accelerator.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-09 in cs.AR).
Categorie arXiv: Hardware Architecture (cs.AR).
Pagina del lavoro: https://arxiv.org/abs/2609.08424 — 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 2026-10-16T13:02:36.888Z dalla pagina https://arxiv.org/catchup/cs.AR/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).