This month's
ASKAP
update
includes discussion of the first ASKAP continuum pipeline output from Setonix,
the first call for observing targets corresponding to the full survey time allocation,
ongoing consolidation efforts, and a brief review of ASKAP at the Astronomical
Society of Australia Annual Science Meeting.
Commissioning of ASKAPsoft and the processing pipeline on the new
Setonix supercomputer has progressed to the point where full continuum
imaging jobs can be run. After making some changes to accommodate the
structure of Setonix (with 128 cores per node and the ability to
assign independent jobs to different cores within a single node),
a full 36-beam observation with 288 x 1 MHz channels can be processed on 72
of the 180 nodes available for ASKAP Operations. For a 10-hour
observation, end-to-end processing takes under 5 hours, easily
allowing us to keep up with incoming data. With additional
optimisation we should be able to achieve the same result on 41 nodes.
The figure above is an ASKAP pipeline job
statistics plot, showing the execution time of various processes
throughout the duration of a full continuum imaging run. (Image credit: Wasim Raja)
|