Akceleracja obliczen algebry liniowej z wykorzystaniem masywnie rownoleglych, wielordzeniowych procesorow GPU

TL;DR

This paper discusses leveraging modern CUDA-supported GPU accelerators for high-performance linear algebra computations, demonstrating over a thousandfold speedup in matrix multiplication compared to CPUs.

Contribution

It explores the use of GPU architecture for linear algebra, showing significant performance improvements and reduced latency in matrix operations.

Findings

Over thousandfold speedup in matrix multiplication

Significant reduction in computation latency

Effective utilization of CUDA-enabled GPUs for linear algebra

Abstract

The paper presents the aspect of use of modern graphics accelerators supporting CUDA technology for high-performance computing in the field of linear algebra. Fully programmable graphic cards have been available for several years for both ordinary users and research units. They provide the capability of performing virtually any computing with high performance, which is often beyond the reach of conventional CPUs. GPU architecture, also in case of classical problems of linear algebra which is the basis for many calculations, can bring many benefits to the developer. Performance increase, observed during matrix multiplication on nVidia Tesla C2050, was more than thousandfold compared to ordinary CPU, resulting in drastic reduction of latency for some of the results, thus the cost of obtaining them.