BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.adass.org//adass2026//talk//QL8JSB
BEGIN:VTIMEZONE
TZID:AWST
BEGIN:STANDARD
DTSTART:20000101T000000
RRULE:FREQ=YEARLY;BYMONTH=1;UNTIL=20051231T160000Z
TZNAME:AWST
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
END:STANDARD
BEGIN:STANDARD
DTSTART:20070325T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:AWST
TZOFFSETFROM:+0900
TZOFFSETTO:+0800
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20071028T030000
RRULE:FREQ=YEARLY;BYDAY=4SU;BYMONTH=10;UNTIL=20081025T190000Z
TZNAME:AWDT
TZOFFSETFROM:+0800
TZOFFSETTO:+0900
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-adass2026-QL8JSB@pretalx.adass.org
DTSTART;TZID=AWST:20261105T114500
DTEND;TZID=AWST:20261105T120000
DESCRIPTION:As part of the EU-funded RADIOBLOCKS project\, we have created 
 a set of highly optimized GPU libraries which can serve as building blocks
  for a correlator or beamformer application. We have integrated a number o
 f these radioblocks into a fully functional VLBI correlator\, which is int
 ended to become the next production correlator of the European VLBI Networ
 k (EVN). In this talk I will present this correlator and its constituent r
 adioblocks.\n\nSince 2012\, the operational correlator of the EVN is a dis
 tributed CPU based correlator which is deployed locally on an HPC cluster.
  However\, the advent of next generation wideband receiver systems is expe
 cted to increase the maximum observed bandwidth in EVN observations by a f
 actor four to eight\, which will require an increase in correlator capacit
 y by a similar amount. Expanding the current HPC cluster to meet these new
  requirements would be prohibitively expensive while also using a signific
 ant amount of power.  However\, as I will show in this talk\, these requir
 ements can be met cost-effectively using GPU acceleration\, while drawing 
 significantly less power. In the talk I will present detailed measurements
  of the power consumption of our GPU implementation.\n\nThe GPU correlator
  is not a monolithic application\; it makes use of a collection of GPU lib
 raries that we have created\, which we call radioblocks. Each of these rad
 ioblocks is distributed as a stand-alone library\, allowing other projects
  to easily incorporate the relevant radioblocks into their own code base.\
 nFor example\, a core component of a VLBI correlator is the geometric dela
 y compensation for which we have created an optimized VLBI capable radiobl
 ock. The (cross-)correlations are performed using a radioblock called the 
 tensor core library (TCC)\, which uses tensor core instructions to compute
  the correlation function. It achieves an order-of-magnitude performance i
 mprovement over non-tensor-core GPU implementations.\n\nFurthermore\, we h
 ave implemented a number of advanced features as radioblocks. We have crea
 ted a radioblock which can perform coherent de-dispersion on baseband data
 \, this feature is crucial to process certain classes of radio transients 
 such as Fast Radio Bursts (FRBs)\, and some pulsars. Another example is a 
 radioblock which can produce multiple simultaneous phase centres\, a techn
 ique to efficiently perform wide-field VLBI. It does this by producing ind
 ividual narrow-field data sets for each source in the field of view\, rath
 er than producing a single monolithic wide-field dataset that encompasses 
 the entire field of view.\n\nI will conclude the talk by discussing a pote
 ntial future application. The Space Array initiative\, of which we are mem
 bers\, is preparing a proposal to build a space based VLBI array. Due to b
 andwidth limitations\, the correlation will be performed onboard of the sp
 acecrafts. I will show how our work could be applied to this space-based c
 orrelator.
DTSTAMP:20261001T111258Z
LOCATION:Banquet Hall
SUMMARY:SFXGPU: A GPU accelerated correlator for the European VLBI Network.
  - Aard Keimpema
URL:https://pretalx.adass.org/adass2026/talk/QL8JSB/
END:VEVENT
END:VCALENDAR