May 18 – 22, 2026
Virginia Tech
America/New_York timezone

Analysis of Floating-Point Matrix Multiplication Computed via Integer Arithmetic

May 19, 2026, 11:50 AM
25m
Torgersen Hall 3100 (Virginia Tech)

Torgersen Hall 3100

Virginia Tech

Minisymposium Talk Approximate Computing in Numerical Linear Algebra Approximate Computing in Numerical Linear Algebra

Speaker

Mantas Mikaitis (University of Leeds)

Description

Ootomo, Ozaki, and Yokota [Int. J. High Perform. Comput. Appl., 38 (2024), p. 297–313] have proposed a strategy to recast a floating-point matrix multiplication in terms of integer matrix products. The factors $A$ and $B$ are split into integer slices, the product of these slices is computed exactly, and $AB$ is approximated by accumulating these integer products in floating-point arithmetic. This technique is particularly well suited to mixed-precision matrix multiply–accumulate units with integer support, such as the NVIDIA tensor cores or the AMD matrix cores The number of slices allows for performance-accuracy tradeoffs: more slices yield better accuracy but require more multiplications, which in turn reduce performance. We propose an inexpensive way to estimate the minimum number of multiplications needed to achieve a prescribed level of accuracy. Our error analysis shows that the algorithm may become inaccurate (or inefficient) if rows of $A$ or columns of $B$ are badly scaled. We perform a range of numerical experiments, both in simulation and on the latest NVIDIA GPUs, that confirm the analysis and illustrate strengths and weaknesses of the algorithm.

Authors

Dr Ahmad Abdelfattah (University of Tennessee) Prof. Jack Dongarra (University of Tennessee) Massimiliano Fasi (University of Leeds) Mantas Mikaitis (University of Leeds) Francoise Tisseur (The University of Manchester)

Presentation materials

There are no materials yet.