DAQHLT TDAQ 1 Nuove release tdaq 02 00

DAQ/HLT

TDAQ (1) • Nuove release tdaq – 02 -00 -00 a dicembre – Altra release prima del run?

TDAQ (2) • Problems and improvements – Ro. I builder • Can use only round-robin – Stops DAQ if a L 2 SVR not responding – ROS • Elonex is no more there… • Tests on new PCs indicate performance increase of ~30% – p. ROS • Uses clears (why)? – Dropped @ high rate • Single-threaded • Garbage collection misbehaving • Increase parallelism – L 2 SVR • Single-threaded – L 2 • Various problems – No clean-up

TDAQ (3) • Problem and improvements – EB • Parallel running partitions can interfere • Run with TCP – SFO • Replacements to be found (substitute Elonex PC) – Mu. Cal • Left-over buffer from L 2 • Left-over L 2 processes • Mu. Cal becomes a “buffer-server” – Event monitoring out of EFD • Based on L 2 and EF decisions • Switch-based ROS – Long term – Due to high request rate from L 2

HLT • Complete review of menus for 10**31 – Justification of menu elements – Efficiency calculation tools – Algorithm stability • Operation – Trigger presenter reviewed – Tools for rate calculation • Monitoring – CAF resources and CERN Tier-3 for trigger monitoring – Online monitoring improvements • Trigger configuration – Currently not user-friendly – Big improvements on trigger tool • Plan for first beams – Special runs with “Enhanced MB” events – Trigger result analysis

Piano finanziario • Approvato a settembre: – 650 k. CHF (406 keuro) • Dopo il problema di LHC – Due variabili: • Evoluzione LHC (Ottobre? ) • Evoluzione tecnologica – Nuove tecnologie (previste per la primavera 2009) » Ritardate per crisi finanziaria?

Possibile piano • In evoluzione • Preparazione infrastruttura 2009 – Preparazione rack • • Servizi Server Switch Incide per il 20% – Preparazione macchine • Vecchia tecnologia commissioning nel run • Nuova tecnologia gara conveniente ? • Contributo previsto “infrastruttura” • 206 keuro, 150 s. j. , 50 al 2010 • Ipotesi ulteriore – 160 al 2010 – Come dividere il resto?