Quasi-real-time analysis of dynamic near field scattering data using a graphics processing unit

TL;DR

This paper demonstrates a GPU-based implementation for analyzing dynamic near field scattering data, significantly reducing processing time from hours to minutes and enabling near real-time analysis across various methods.

Contribution

The paper introduces an optimized GPU data management scheme that accelerates dynamic NFS data analysis, making it feasible in real-time or near real-time conditions.

Findings

Processing time reduced from hours to minutes

Applicable to multiple NFS methods

Processing now comparable to data acquisition time

Abstract

We present an implementation of the analysis of dynamic near field scattering (NFS) data using a graphics processing unit (GPU). We introduce an optimized data management scheme thereby limiting the number of operations required. Overall, we reduce the processing time from hours to minutes, for typical experimental conditions. Previously the limiting step in such experiments, the processing time is now comparable to the data acquisition time. Our approach is applicable to various dynamic NFS methods, including shadowgraph, Schlieren and differential dynamic microscopy.