What type of study is this?

This is a Quantitative Study study.

October 3, 2025Open Access

High Performance Matrix Multiplication

Key Points

CuBLAS outperforms other algorithms for large square matrices, with a statistically significant p-value below 5e-12.
The study compares five matrix multiplication algorithms, assessing their performance using floating point operations per second.
Matrix multiplication serves as a cornerstone for high performance applications like deep learning and scientific simulations.
The algorithms analyzed include CuBLAS, CUDA, BLAS, OpenMP, and C++ Threads, evaluated with square matrices over 10,000 in size.

Abstract

Matrix multiplication is the foundation from much of the success from high performance technologies like deep learning, scientific simulations, and video graphics. High level programming languages like Python and R rely on highly optimized low level libraries for performing core linear algebra operations like matrix multiplication from Basic Linear Algebra Subprograms (BLAS). This paper compares the performance of five different matrix multiplication algorithms using CuBLAS, CUDA, BLAS, OpenMP, and C++ Threads. We find statistical significance with a p-value below 5e-12 to support the hypothesis that for square N N matrices where N is at least 10, 000 then the in order performance as measured in floating point operations per second (FLOPS) for these matrix multiplication algorithms is CuBLAS, CUDA, BLAS, OpenMP, and C++ Threads.

Read Full Paperexternally

Bookmark

View Full Paper

Cite This Study

Ellery W. Davis (Thu,) studied this question.

synapsesocial.com/papers/68e02f46f0e39f13e7fa2c6e https://doi.org/https://doi.org/10.48550/arxiv.2509.04594

Bookmark

View Full Paper