Exact Fused Dot Product Add Operators - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Exact Fused Dot Product Add Operators

Résumé

This article explores architectures of exact (correctly rounded) fused dot product and add operators suitable for the FP32 and FP64 binary floating-point representations with subnormal support, and other representations with a wide dynamic range such as bfloat16. The exact summation of terms before rounding requires a full-size accumulator, and this work discusses techniques to compress the identical bits of this accumulator. This requires the computation of the relative shift amounts of the terms, which is formulated as a parallel prefix algorithm, allowing for a low-latency implementation. Architectural options for the exact fused dot product and add operators with up to 16 products for FP32, FP64 and mixed-precision BF16 to FP32 are evaluated using the TSMC 16FFC technology node.
Fichier principal
Vignette du fichier
ARITH2023.pdf (336.9 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04240762 , version 1 (13-10-2023)

Licence

Paternité

Identifiants

  • HAL Id : hal-04240762 , version 1

Citer

Orégane Desrentes, Benoît Dupont de Dinechin, Florent de Dinechin. Exact Fused Dot Product Add Operators. 2023 ARITH - 30th IEEE International Symposium on Computer Arithmetic, Sep 2023, Portland, OR, United States. ⟨hal-04240762⟩
59 Consultations
69 Téléchargements

Partager

Gmail Facebook X LinkedIn More