FTTN: Feature-Targeted Testing for Numerical Properties of NVIDIA & AMD Matrix Accelerators

RESOURCE

Abstract

FTTN is a test suite to evaluate the numerical behaviors of matrix accelerators of GPUs (NVIDIA Tensor Cores and AMD Matrix Cores) in a quick and simple setting. Matrix accelerators are heavily used in today's computationally intense applications to speed up matrix multiplications. This test suite provides a comprehensive study on the numerical behaviors of these accelerators, including support for subnormals, rounding modes, extra precision bits and FMA features. Is there
Developers:
Laguna Peralta, Ignacio [1] Gopalakrishnan, Ganesh [2] Li, Xinyi [2]
  1. Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
  2. University of Utah
Release Date:
2024-08-28
Project Type:
Open Source, Publicly Available Repository
Software Type:
Scientific
Version:
0.1
Licenses:
MIT License
Sponsoring Org.:
Code ID:
145775
Site Accession Number:
LLNL-CODE-2000523
Research Org.:
Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
Country of Origin:
United States

RESOURCE

Citation Formats

Laguna Peralta, Ignacio, Gopalakrishnan, Ganesh, and Li, Xinyi. FTTN: Feature-Targeted Testing for Numerical Properties of NVIDIA & AMD Matrix Accelerators. Computer Software. https://github.com/LLNL/FTTN. USDOE National Nuclear Security Administration (NNSA). 28 Aug. 2024. Web. doi:10.11578/dc.20241018.1.
Laguna Peralta, Ignacio, Gopalakrishnan, Ganesh, & Li, Xinyi. (2024, August 28). FTTN: Feature-Targeted Testing for Numerical Properties of NVIDIA & AMD Matrix Accelerators. [Computer software]. https://github.com/LLNL/FTTN. https://doi.org/10.11578/dc.20241018.1.
Laguna Peralta, Ignacio, Gopalakrishnan, Ganesh, and Li, Xinyi. "FTTN: Feature-Targeted Testing for Numerical Properties of NVIDIA & AMD Matrix Accelerators." Computer software. August 28, 2024. https://github.com/LLNL/FTTN. https://doi.org/10.11578/dc.20241018.1.
@misc{ doecode_145775,
title = {FTTN: Feature-Targeted Testing for Numerical Properties of NVIDIA & AMD Matrix Accelerators},
author = {Laguna Peralta, Ignacio and Gopalakrishnan, Ganesh and Li, Xinyi},
abstractNote = {FTTN is a test suite to evaluate the numerical behaviors of matrix accelerators of GPUs (NVIDIA Tensor Cores and AMD Matrix Cores) in a quick and simple setting. Matrix accelerators are heavily used in today's computationally intense applications to speed up matrix multiplications. This test suite provides a comprehensive study on the numerical behaviors of these accelerators, including support for subnormals, rounding modes, extra precision bits and FMA features. Is there},
doi = {10.11578/dc.20241018.1},
url = {https://doi.org/10.11578/dc.20241018.1},
howpublished = {[Computer Software] \url{https://doi.org/10.11578/dc.20241018.1}},
year = {2024},
month = {aug}
}