Abstracting FPGAs in Python for fun and radio astronomy
Room A | Fri 23 Jan 1:30 p.m.–2:15 p.m.
Presented by
-
Andrew is a software engineer at Australia's national science agency, CSIRO. He mostly works on the Square Kilometre Array project, a scientific megaproject currently under construction in both Western Australia and South Africa. Specifically, Andrew writes (open source!) Python software to control, monitor, and test the Correlator and Beamformer for the Low frequency telescope being built in WA. This is an FPGA-based real-time digital signal processing system operating at terabit speeds to process antenna signals for astronomers to use.
Before CSIRO, Andrew worked as an automation/controls engineer in heavy industry. He still automates complex real-time processes, only now they involve digital signal processing instead of steam and turbines.
Andrew is a software engineer at Australia's national science agency, CSIRO. He mostly works on the Square Kilometre Array project, a scientific megaproject currently under construction in both Western Australia and South Africa. Specifically, Andrew writes (open source!) Python software to control, monitor, and test the Correlator and Beamformer for the Low frequency telescope being built in WA. This is an FPGA-based real-time digital signal processing system operating at terabit speeds to process antenna signals for astronomers to use.
Before CSIRO, Andrew worked as an automation/controls engineer in heavy industry. He still automates complex real-time processes, only now they involve digital signal processing instead of steam and turbines.
Abstract
A shiny, generic, "Pythonic" abstraction rather than a bunch of register addresses - a good idea, right?
The SKA-Low radio telescope, part of the international Square Kilometre Array scientific megaproject, is currently under construction in Western Australia. CSIRO is developing the SKA-Low Correlator and Beamformer, which will process 6 Tbps of input signals (from up to 512 antenna "stations") in real time. The Correlator produces data streams that can be used by astronomers to create images, and the Beamformers digitally steer the antenna array to point in specific directions. Together they can produce up to 9 Tbps of scientific data streams.
This real-time signal processing is performed on FPGAs, with Python software for monitoring and control. The SKA-Low telescope is still in the early stages of construction, but the Correlator and Beamformer is already in use, allowing the production of images and the observation of pulsars for commissioning purposes.
This talk will focus on the FPGA-Python interface. Ideas, aspirations, how they compare to reality, and what should have been done differently.
A shiny, generic, "Pythonic" abstraction rather than a bunch of register addresses - a good idea, right?
The SKA-Low radio telescope, part of the international Square Kilometre Array scientific megaproject, is currently under construction in Western Australia. CSIRO is developing the SKA-Low Correlator and Beamformer, which will process 6 Tbps of input signals (from up to 512 antenna "stations") in real time. The Correlator produces data streams that can be used by astronomers to create images, and the Beamformers digitally steer the antenna array to point in specific directions. Together they can produce up to 9 Tbps of scientific data streams.
This real-time signal processing is performed on FPGAs, with Python software for monitoring and control. The SKA-Low telescope is still in the early stages of construction, but the Correlator and Beamformer is already in use, allowing the production of images and the observation of pulsars for commissioning purposes.
This talk will focus on the FPGA-Python interface. Ideas, aspirations, how they compare to reality, and what should have been done differently.