Comparing the performance of different x86 SIMD instruction sets for a medical imaging application on modern multi- and manycore chips

Authors:
Johannes Hofmann;Jan Treibig;Georg Hager;Gerhard Wellein
Affiliations:
University Erlangen-Nuremberg, Erlangen, Germany;University Erlangen-Nuremberg, Erlangen, Germany;University Erlangen-Nuremberg, Erlangen, Germany;University Erlangen-Nuremberg, Erlangen, Germany
Venue:
Proceedings of the 2014 Workshop on Programming models for SIMD/Vector processing
Year:
2014

Citing 3
Cited 0

LIKWID: A Lightweight Performance-Oriented Tool Suite for x86 Multicore Environments

ICPPW '10 Proceedings of the 2010 39th International Conference on Parallel Processing Workshops
Evaluation of state-of-the-art hardware architectures for fast cone-beam CT reconstruction

Parallel Computing
Pushing the limits for medical image reconstruction on recent standard multicore processors

International Journal of High Performance Computing Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

Single Instruction, Multiple Data (SIMD) vectorization is a major driver of performance in current architectures, and is mandatory for achieving good performance with codes that are limited by instruction throughput. We investigate the efficiency of different SIMD-vectorized implementations of the RabbitCT benchmark. RabbitCT performs 3D image reconstruction by back projection, a vital operation in computed tomography applications. The underlying algorithm is a challenge for vectorization because it consists, apart from a streaming part, also of a bilinear interpolation requiring scattered access to image data. We analyze the performance of SSE (128 bit), AVX (256 bit), AVX2 (256 bit), and IMCI (512 bit) implementations on recent Intel x86 systems. A special emphasis is put on the vector gather implementation on Intel Haswell and Knights Corner microarchitectures. Finally we discuss why GPU implementations perform much better for this specific algorithm.