Low-complexity Distributed Tomographic Backprojection for large datasets

TL;DR

This paper introduces a GPU-accelerated, distributed tomographic backprojection algorithm that significantly reduces computational complexity, enabling rapid reconstructions of large datasets from synchrotron sources within seconds.

Contribution

The work presents a novel low-complexity formula-based algorithm and its efficient GPU implementation for large-scale tomographic reconstruction, outperforming traditional methods in speed.

Findings

Reconstruction achieved within 1 second with complete data transfer.

Algorithm operates with O(N^3 log N) complexity, much less than traditional O(N^4).

Effective distribution across 4 GPUs for large datasets.

Abstract

In this manuscript we present a fast GPU implementation for tomographic reconstruction of large datasets using data obtained at the Brazilian synchrotron light source. The algorithm is distributed in a cluster with 4 GPUs through a fast pipeline implemented in C programming language. Our algorithm is theoretically based on a recently discovered low complexity formula, computing the total volume within O(N3logN) floating point operations; much less than traditional algorithms that operates with O(N4) flops over an input data of size O(N3). The results obtained with real data indicate that a reconstruction can be achieved within 1 second provided the data is transferred completely to the memory.