Multiple Double Arithmetic on NVIDIA Tensor Cores

DOI10.48550/arXiv.2607.06881MaRDI QIDQ7284076arXivFDO

Authors Howard Cheng, Howard Chen, Jan Verschelde

Publication date 8 July 2026

Published in arXiv

Tldr We extend double arithmetic operations in NVIDIA Tensor Cores with multiple doubles using the Ozaki scheme, enabling high performance and low memory usage for dense linear algebra applications.

Full work available at URL https://doi.org/10.48550/arXiv.2607.06881

arXiv classification

cs.DCcs.MSmath.NA

Cites work

• A floating-point technique for extending the available precision
• Accelerated Polynomial Evaluation and Differentiation at Power Series in Multiple Double Precision
• Adaptive precision floating-point arithmetic and fast robust geometric predicates
• Algorithm 795
• CAMPARY: CUDA multiple precision arithmetic library and applications
• Error-free transformations of matrix multiplication by using fast routines of matrix multiplication and its applications
• GPU accelerated Newton for Taylor series solutions of polynomial homotopies in multiple double precision
• Least Squares on GPUs in Multiple Double Precision
• Mixed precision algorithms in numerical linear algebra
• Multiword Arithmetic and Parallel Computing
• Verification methods: rigorous results using floating-point arithmetic

This page was built for publication: Multiple Double Arithmetic on NVIDIA Tensor Cores

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284076)