EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

On the Impact of Quantization Precision and Calibration Settings Applied to the SSF Neural Video Decoder

Ruhan Conceição, Wen-Hsiao Peng, Marcelo Porto, Luciano Agostini

Journal of Integrated Circuits and Systems · 12.04.2026 · 21 (1) · 1-10

DOI: 10.29292/jics.v21i1.1086 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Integrated Circuits and Systems, online erschienen am 12.04.2026. Verfasst von Ruhan Conceição und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Neural video codecs have recently achieved state-of-the-art performance, but their reliance on floating-point arithmetic introduces cross-platform inconsistencies. In this work, we present an extended investigation of post-training quantization (PTQ) applied to the hyper prior decoding path in the Scale-Space Flow (SSF) codec, focusing on its role in ensuring cross-platform consistency while maintaining compression efficiency. We systematically evaluate the impact of different quantization bit-depths (4-, 8-, 12-, and 16-bit integers), calibration configurations (crop sizes and dataset sizes), and temporal test scenarios (short and long P-frame chains). The results show that aggressive quantization (INT4) severely degrades rate-distortion (RD) performance, whereas INT8 achieves a favorable trade-off with BD-Rate penalties below 7%. Furthermore, INT12 precision is sufficient to match the FP32 baseline with virtually no loss. Calibration analysis reveals that high-precision quantization is only marginally affected by dataset size and resolution, while lower-precision scenarios exhibit less predictable behavior. Finally, an estimation of energy consumption for a dedicated hardware design indicates that quantization can reduce energy consumption by up to 84% compared to the FP32 anchor model, while preserving near-floating-point efficiency.

Zitieren

APA

Conceição, R., Peng, W., Porto, M., & Agostini, L. (2026). On the Impact of Quantization Precision and Calibration Settings Applied to the SSF Neural Video Decoder. Journal of Integrated Circuits and Systems, 21(1), 1-10. https://doi.org/10.29292/jics.v21i1.1086

Vancouver

Conceição R, Peng W, Porto M, Agostini L. On the Impact of Quantization Precision and Calibration Settings Applied to the SSF Neural Video Decoder. Journal of Integrated Circuits and Systems. 2026;21(1):1-10. doi:10.29292/jics.v21i1.1086

BibTeX
@article{Conceio2026,
  author = {Conceição, Ruhan and Peng, Wen-Hsiao and Porto, Marcelo and Agostini, Luciano},
  title = {On the Impact of Quantization Precision and Calibration Settings Applied to the SSF Neural Video Decoder},
  journal = {Journal of Integrated Circuits and Systems},
  year = {2026},
  volume = {21},
  number = {1},
  pages = {1--10},
  doi = {10.29292/jics.v21i1.1086}
}
Kostenlos starten →