8 August 2016 Status report of the end-to-end ASKAP software system: towards early science operations
Author Affiliations +
Abstract
The Australian SKA Pathfinder (ASKAP) is a novel centimetre radio synthesis telescope currently in the commissioning phase and located in the midwest region of Western Australia. It comprises of 36 x 12 m diameter reflector antennas each equipped with state-of-the-art and award winning Phased Array Feeds (PAF) technology. The PAFs provide a wide, 30 square degree field-of-view by forming up to 36 separate dual-polarisation beams at once. This results in a high data rate: 70 TB of correlated visibilities in an 8-hour observation, requiring custom-written, high-performance software running in dedicated High Performance Computing (HPC) facilities. The first six antennas equipped with first-generation PAF technology (Mark I), named the Boolardy Engineering Test Array (BETA) have been in use since 2014 as a platform to test PAF calibration and imaging techniques, and along the way it has been producing some great science results. Commissioning of the ASKAP Array Release 1, that is the first six antennas with second-generation PAFs (Mark II) is currently under way. An integral part of the instrument is the Central Processor platform hosted at the Pawsey Supercomputing Centre in Perth, which executes custom-written software pipelines, designed specifically to meet the ASKAP imaging requirements of wide field of view and high dynamic range. There are three key hardware components of the Central Processor: The ingest nodes (16 x node cluster), the fast temporary storage (1 PB Lustre file system) and the processing supercomputer (200 TFlop system). This High-Performance Computing (HPC) platform is managed and supported by the Pawsey support team. Due to the limited amount of data generated by BETA and the first ASKAP Array Release, the Central Processor platform has been running in a more “traditional” or user-interactive mode. But this is about to change: integration and verification of the online ingest pipeline starts in early 2016, which is required to support the full 300 MHz bandwidth for Array Release 1; followed by the deployment of the real-time data processing components. In addition to the Central Processor, the first production release of the CSIRO ASKAP Science Data Archive (CASDA) has also been deployed in one of the Pawsey Supercomputing Centre facilities and it is integrated to the end-to-end ASKAP data flow system. This paper describes the current status of the “end-to-end” data flow software system from preparing observations to data acquisition, processing and archiving; and the challenges of integrating an HPC facility as a key part of the instrument. It also shares some lessons learned since the start of integration activities and the challenges ahead in preparation for the start of the Early Science program.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Juan Carlos Guzman, Juan Carlos Guzman, Jessica Chapman, Jessica Chapman, Malte Marquarding, Malte Marquarding, Matthew Whiting, Matthew Whiting, } "Status report of the end-to-end ASKAP software system: towards early science operations", Proc. SPIE 9913, Software and Cyberinfrastructure for Astronomy IV, 991311 (8 August 2016); doi: 10.1117/12.2232513; https://doi.org/10.1117/12.2232513
PROCEEDINGS
12 PAGES


SHARE
RELATED CONTENT

The human pipeline: distributed data reduction for ALMA
Proceedings of SPIE (August 06 2014)
ALMA observations during its first early science cycles
Proceedings of SPIE (August 06 2014)
Square Kilometer Array project status report
Proceedings of SPIE (July 13 2018)
Data products of the ALMA and NRAO archives
Proceedings of SPIE (July 15 2016)
Data management for the EVLA
Proceedings of SPIE (September 24 2012)

Back to Top