Development of DiFX Adam Deller NRAO 3rd DiFX workshop, Curtin University, Perth
Adam Deller3rd DiFX workshop, Curtin University, Perth Outline Reasons for the development of DiFX Development of fxcorr (pre-DiFX) Rationale and planning of DiFX DiFX for the LBA DiFX at NRAO and elsewhere
Adam Deller3rd DiFX workshop, Curtin University, Perth DiFX: the raison d’etre In 2004, the LBA installed new disk-based recorders capable of 4x the bandwidth of the S2 tape system The existing hardware correlator could not interface to the new recorders Required: a “stop-gap” solution before the CABB correlator for ATCA was ready for VLBI data The S2 correlator
Adam Deller3rd DiFX workshop, Curtin University, Perth DiFX: the raison d’etre
Adam Deller3rd DiFX workshop, Curtin University, Perth 2005: Development of fxcorr My PhD began February first attempt was FX-style correlator which ran on a single node (distribution via scripts/data division) Naivete in both C++ and correlators led me to build in some idiosyncracies/inefficiencies Verified against S2 results, but plans made to redesign from scratch, using data distribution with MPI - mpifxcorr, copy/pasting the internals of the correlation code
Adam Deller3rd DiFX workshop, Curtin University, Perth Rationale/planning of DiFX At this stage, plans made to support other formats than LBA (specifically Mark5) Attempted to not “design out” any general applications, but the main focus was nevertheless on LBA operations Also made provisions from the outset for eVLBI connections into the correlator Main development was complete by mid 2006
Adam Deller3rd DiFX workshop, Curtin University, Perth DiFX at the LBA DiFX has been the production correlator of the LBA since end of 2006 Plans to use the CABB correlator as the LBA correlator have been shelved 2008: Correlation moves from Swinburne to Curtin Oversubscription rate has risen substantially since DiFX + LBADR (recorders) introduced - maybe ~1.5 to ~3??
Adam Deller3rd DiFX workshop, Curtin University, Perth Science highlights at the LBA First trans-Tasman fringes - to a 6m dish in NZ!
Adam Deller3rd DiFX workshop, Curtin University, Perth Science highlights at the LBA First trans-Tasman fringes - to a 6m dish in NZ! First southern hemisphere eVLBI, detecting a flare in Circinus X1
Adam Deller3rd DiFX workshop, Curtin University, Perth Science highlights at the LBA First trans-Tasman fringes - to a 6m dish in NZ! First southern hemisphere eVLBI, detecting a flare in Circinus X1 6 new pulsar parallaxes, trebling the number of Southern pulsar parallaxes
Adam Deller3rd DiFX workshop, Curtin University, Perth DiFX at the VLBA VLBA sensitivity upgrade has been an ongoing project that requires a new correlator to deal with increased data rates DiFX selected in 2008 Going “live” this proposal cycle - first element of the upgrade (backends and recorders to come) to do so Huge amounts of streamlining and infrastructure have been contributed by the NRAO efforts
Adam Deller3rd DiFX workshop, Curtin University, Perth DiFX worldwide DiFX is likely to replace three other existing hardware correlators: Bonn, USNO, Haystack This is the third DiFX workshop - attendance has gone from about 8 to about 12 to about 24 - the last year especially has been an exciting time development- and adoption- wise Lets keep this thing rolling!