Parallel GRB Source Localization Pipelines for the Advanced Particle-Astrophysics Telescope
SessionResearch Posters Display
Event Type
Posters
Research Posters
TP
XO / EX
TimeTuesday, 16 November 20218:30am - 5pm CST
LocationSecond Floor Atrium
DescriptionThe Advanced Particle-astrophysics Telescope (APT) is a planned space-based observatory to survey the entire sky for gamma-ray bursts (GRBs). It seeks to promptly detect these transient events, then communicate with narrow-band instruments for follow-up observations. To this end, we are developing analytical methods for real-time detection and localization of GRBs, targeting computing hardware that might fly onboard the orbiting platform.
In prior work, we implemented a three-stage software pipeline on a low-powered ARM CPU. In this work, we present GPU-accelerated algorithms for the last two pipeline stages, measure execution times on an NVIDIA GeForce RTX 2080, then apply a scaling factor to estimate performance on a 10-watt NVIDIA Jetson NX Xavier board. For the second stage, we demonstrate an estimated speedup of 3-3.5x over the CPU implementation for large datasets. For the third stage, we demonstrate that GPU acceleration significantly slows the increase of execution time with input size.
In prior work, we implemented a three-stage software pipeline on a low-powered ARM CPU. In this work, we present GPU-accelerated algorithms for the last two pipeline stages, measure execution times on an NVIDIA GeForce RTX 2080, then apply a scaling factor to estimate performance on a 10-watt NVIDIA Jetson NX Xavier board. For the second stage, we demonstrate an estimated speedup of 3-3.5x over the CPU implementation for large datasets. For the third stage, we demonstrate that GPU acceleration significantly slows the increase of execution time with input size.
Archive view
