cuFFS: A GPU-accelerated code for Fast Faraday Rotation Measure Synthesis

© Sridhar, Heald & van der Hulst (2018)

Rotation measure (RM) synthesis is a widely used polarization processing algorithm for reconstructing polarized structures along the line of sight. Performing RM synthesis on large datasets produced by telescopes like LOFAR can be computationally intensive. The required computational cost is likely to get worse due to the planned large area sky surveys with telescopes like the Low Frequency Array (LOFAR), the Murchison Widefield Array (MWA), and eventually the Square Kilometre Array (SKA).

The massively parallel General Purpose Graphical Processing Units (GPGPUs) can be used to execute some of the computationally intensive astronomical image processing algorithms including RM synthesis. In this paper, we present a GPU-accelerated code, called cuFFS or CUDA-accelerated Fast Faraday Synthesis, to perform Faraday rotation measure synthesis. Compared to a fast single-threaded and vectorized CPU implementation, depending on the structure and format of the data cubes, our code achieves an increase in speed of up to two orders of magnitude. The code is written in C and supports both FITS and HDF5 file formats. The code is available at https://github.com/sarrvesh/cuFFS .

The image on the left shows a single Faraday depth plane computed with cuFFS using data from the WSRT SING survey (Heald et al. 2009). The plot on the right shows the pixel-by-pixel comparison of the cuFFS output with a CPU implementation.

The paper describing cuFFS has been accepted for publication with Astronomy & Computing and the preprint is now available online ( http://arxiv.org/abs/1810.06250 ).