Petascale XCT: 3D Image Reconstruction with Hierarchical Communications on Multi-GPU Nodes

TL;DR

This paper presents a high-performance system for large-scale 3D X-ray tomography reconstruction on multi-GPU clusters, achieving rapid processing of terabyte-scale volumes by novel optimizations and hierarchical communication strategies.

Contribution

It introduces three novel optimizations for iterative 3D reconstruction algorithms, enabling efficient scaling on supercomputers for petascale datasets.

Findings

Reconstructed a mouse brain volume in under three minutes.

Achieved 65 PFLOPS, 34% of Summit's peak performance.

Demonstrated scalability on thousands of GPUs.

Abstract

X-ray computed tomography is a commonly used technique for noninvasive imaging at synchrotron facilities. Iterative tomographic reconstruction algorithms are often preferred for recovering high quality 3D volumetric images from 2D X-ray images, however, their use has been limited to small/medium datasets due to their computational requirements. In this paper, we propose a high-performance iterative reconstruction system for terabyte(s)-scale 3D volumes. Our design involves three novel optimizations: (1) optimization of (back)projection operators by extending the 2D memory-centric approach to 3D; (2) performing hierarchical communications by exploiting "fat-node" architecture with many GPUs; (3) utilization of mixed-precision types while preserving convergence rate and quality. We extensively evaluate the proposed optimizations and scaling on the Summit supercomputer. Our largest reconstruction is a mouse brain volume with 9Kx11Kx11K voxels, where the total reconstruction time is under three minutes using 24,576 GPUs, reaching 65 PFLOPS: 34% of Summit's peak performance.