Hardware, software, hybrid correlators

1
Hardware, software, hybrid correlators

• Panel Discussion
• Groningen, June 28

2
Panel members

• Brent Carlson
• Andre Gunst
• Peter Hall (chair)
• Tetsuro Kondo
• Robert Navarro
• Kjeld van der Schaaf
• Martijn van Veelen

3
Definitions

• Hardware
• ASIC or FPGA based system
• Purpose built or re-usable (Werthimer)
• Operating software primarily from h/w engineering
  world
• Software
• General purpose or customized computer
• Large software component can be from system
  literate astronomers s/w engineers
• Hybrid
• Computer with external hardware engines
• Both h/w engineering and other software
  contributions

4
Some starting observations

• There is no magic
• Very big, next generation (SKA), DSP systems will
  be expensive and require large development teams.
  
• SKA scale feasible operating re-investment
  budget lead us to consider merits of alternative
  approaches
• Refer earlier SKA Ref Design talk
• The SKA pace available astronomy,
  astro-engineering community numbers forces us to
  have vastly increased industry engagement
• How much should DSP subsystems figure in this?
• There is life outside the SKA
• The optimum SKA approach may not translate to
  e.g. a campus RA experiment
• Even SKA will have data spigots to support
  innovation
• In comparing correlator approaches, need to
  consider performance both fatal flaw and net
  present value analyses

5
Comparisons

• Martijn van Veelen

6
Factors For Next Generation Systems

• Hardware
• Software
• Hybrid
• Sociological