1 April 2020 in The Astrophysical Journal Supplement Series
The detection of non-repeating or irregular events in time-domain radio astronomy has gained importance over the last decade due to the discovery of fast radio bursts. Existing or upcoming radio telescopes are gathering more and more data, and consequently, the software, which is an important part of these telescopes, must process large data volumes at high data rates. Data has to be searched through to detect new and interesting events, often in real time. These requirements necessitate new and fast algorithms that must process data quickly and accurately. In this work, we present new algorithms for single-pulse detection using boxcar filters. We have quantified the signal loss introduced by single-pulse-detection algorithms, which use boxcar filters, and based on these results, we have designed two distinct “lossy” algorithms. Our lossy algorithms use an incomplete set of boxcar filters to accelerate detection at the expense of a small reduction in detected signal power. We present formulae for signal loss, descriptions of our algorithms and their parallel implementation on NVIDIA GPUs using CUDA. We also present tests of correctness, tests on artificial data, and the performance achieved. Our implementation can process SKA-MID-like data 266× faster than real time on an NVIDIA P100 GPU and 500× faster than real time on an NVIDIA Titan V GPU with a mean signal power loss of 7%. We conclude with prospects for single-pulse detection for beyond the SKA era, nanosecond time-resolution radio astronomy.
1 December 2018 in The Astrophysical Journal Supplement Series
The study of binary pulsars enables tests of general relativity. Orbital motion in binary systems causes the apparent pulsar spin frequency to drift, reducing the sensitivity of periodicity searches. Acceleration searches are methods that account for the effect of orbital acceleration. Existing methods are currently computationally expensive, and the vast amount of data that will be produced by next-generation instruments such as the Square Kilometre Array necessitates real-time acceleration searches, which in turn requires the use of high-performance computing (HPC) platforms. We present our implementation of the correlation technique for the Fourier Domain Acceleration Search (FDAS) algorithm on Graphics Processor Units (GPUs). The correlation technique is applied as a convolution with multiple finite impulse response (FIR) filters in the Fourier domain. Two approaches are compared: the first uses the NVIDIA cuFFT library for applying Fast Fourier transforms (FFTs) on the GPU, and the second contains a custom FFT implementation in GPU shared memory. We find that the FFT shared-memory implementation performs between 1.5 and 3.2 times faster than our cuFFT-based application for smaller but sufficient filter sizes. It is also 4–6 times faster than the existing GPU and OpenMP implementations of FDAS. This work is part of the AstroAccelerate project, a many-core accelerated time-domain signal-processing library for radio astronomy.
21 January 2021 in IEEE Access
The Square Kilometer Array (SKA) is an international initiative for developing the world's largest radio telescope with a total collecting area of over a million square meters. The scale of the operation, combined with the remote location of the telescope, requires the use of energy-efficient computational algorithms. This, along with the extreme data rates that will be produced by the SKA and the requirement for a real-time observing capability, necessitates in-situ data processing in an edge style computing solution. More generally, energy efficiency in the modern computing landscape is becoming of paramount concern. Whether it be the power budget that can limit some of the world's largest supercomputers, or the limited power available to the smallest Internet-of-Things devices. In this article, we study the impact of hardware frequency scaling on the energy consumption and execution time of the Fast Fourier Transform (FFT) on NVIDIA GPUs using the cuFFT library. The FFT is used in many areas of science and it is one of the key algorithms used in radio astronomy data processing pipelines. Through the use of frequency scaling, we show that we can lower the power consumption of the NVIDIA A100 GPU when computing the FFT by up to 47% compared to the boost clock frequency, with less than a 10% increase in the execution time. Furthermore, using one common core clock frequency for all tested FFT lengths, we show on average a 43% reduction in power consumption compared to the boost core clock frequency with an increase in the execution time still below 10%. We demonstrate how these results can be used to lower the power consumption of existing data processing pipelines. These savings, when considered over years of operation, can yield significant financial savings, but can also lead to a significant reduction of greenhouse gas emissions.
3 August 2020 in ACM Transactions on Architecture and Code Optimization
We present an implementation of the overlap-and-save method, a method for the convolution of very long signals with short response functions, which is tailored to GPUs. We have implemented several FFT algorithms (using the CUDA programming language), which exploit GPU shared memory, allowing for GPU accelerated convolution. We compare our implementation with an implementation of the overlap-and-save algorithm utilizing the NVIDIA FFT library (cuFFT). We demonstrate that by using a shared-memory-based FFT, we can achieved significant speed-ups for certain problem sizes and lower the memory requirements of the overlap-and-save method on GPUs.
27 September 2013 in Classical and Quantum Gravity
20 December 2016 in The Astrophysical Journal
Twin-peak quasi-periodic oscillations (QPOs) are observed in the X-ray power-density spectra of several accreting low-mass neutron star (NS) binaries. In our previous work we have considered several QPO models. We have identified and explored mass–angular-momentum relations implied by individual QPO models for the atoll source 4U 1636-53. In this paper we extend our study and confront QPO models with various NS equations of state (EoS). We start with simplified calculations assuming Kerr background geometry and then present results of detailed calculations considering the influence of NS quadrupole moment (related to rotationally induced NS oblateness) assuming Hartle–Thorne spacetimes. We show that the application of concrete EoS together with a particular QPO model yields a specific mass–angular-momentum relation. However, we demonstrate that the degeneracy in mass and angular momentum can be removed when the NS spin frequency inferred from the X-ray burst observations is considered. We inspect a large set of EoS and discuss their compatibility with the considered QPO models. We conclude that when the NS spin frequency in 4U 1636-53 is close to 580 Hz, we can exclude 51 of the 90 considered combinations of EoS and QPO models. We also discuss additional restrictions that may exclude even more combinations. Namely, 13 EOS are compatible with the observed twin-peak QPOs and the relativistic precession model. However, when considering the low-frequency QPOs and Lense–Thirring precession, only 5 EOS are compatible with the model.