Overview remarks TDAQ Working group lamentations summary hope

Overview, remarks, TDAQ Working group lamentations, summary hope and despair M. Sozzi Liverpool – August 30 th 2013

Outline • Common TDAQ system (TDCB, TEL 62) JTAG testing of TEL 62 (N. De Simone) LAV findings (V. Kozuharov) TEL 62 measurements (R. Piandani) PATTI board (G. Lamanna) • L 0 trigger LKr trigger status (A. Salamon) LKr trigger receiver card (B. Checcucci) RICH trigger (M. Piccini) • L 0 TP Status (E. Gamberini) Status (D. Soldi) • GPU triggering RICH L 0 (E. Graverini) Straws L 1 (J. Pinzino) Data transfer (G. Lamanna) • Challenges and plan + my views on something

1. Common TDAQ system

TDCB Final production started All PCBs available Mounting to start in September Cables ordered for most detectors

TEL 62 JTAG testing

TEL 62 – HW status CEDAR: 1, CHANTI: 1, LAV: 3, STRAWS: 1, RICH: 1, LKR/L 0: 3, MUV: 2 Out of 15 pre-production boards: - 6 had or developed problems, at least 5 of them due to mounting (bad FPGA soldering), several pins - 1 stopped working @CERN after a power cycle: broken protection component on crate reset line (again)

PATTI for TDCB/TEL 62 testing

~ 800 KEUR/board (+ cables)

TEL 62 firmware “Version 1”: used in dry run Upgraded version of last year’s technical run, available on web Known rate limitations Some new features added (e. g. EOB counters) “Version 2”: ready now, under debugging Completely changed DDR memory usage: much faster & no limitation on number of slots All which follows was done with Version 1

TEL 62 performance (dry run) Several (unexpected) problems and low sustainable rate, due to very different causes: - Hardware failures - Configuration errors - Missing protection against unexperienced effects (‘ 0’ triggers, 217 events, …) - Undiscovered firmware bugs - Shallow diagnostics on a quite complex system - Mechanical issues

LAV findings

Lab investigations (V 1)

Expected improvement with V 2

Hard to explain with TEL 62 behaviour

My views on it We now experience the result of known manpower problems TEL 62 system is still fragile (diagnostics, procedures and even mechanics) Lab test capabilities not sufficient: more powerful setup available now: multi-burst, TALK for triggering, PATTI for data on many channels More independent testing and debugging needed also elsewhere Lots of firmware work still ahead of us, hope V 2 firmware allows rate goal We seem to have a reliable hardware testing system Mounting firm to be changed: responsibility on them (higher cost) FPGA upgrade very advisable (higher cost) Plan: - production of all TDCB ongoing - continue testing and debugging - test Version 2 firmware - build 5 new boards (upgraded FPGA and new firm) soon - aiming to start production at end of the year (? ) - board testing in Pisa and Roma Tor Vergata

2. L 0 trigger

Lkr/L 0 trigger

Lkr/L 0 receiver card

My views on it • LKr/L 0 trigger progressing, but still not completely defined nor integrated. Detailed simulation still missing • RICH L 0 trigger FW started, quite some work for integration • Completely missing any MUV L 0 trigger. MUV 1+2 in trigger: feasible at all? • Still missing TEL 62 interconnection (hardware+firmware) for LAV and RICH L 0 trigger: Frascati + Perugia will work on this • Missing CHOD L 0 trigger • Proposal for a straws L 0: unclear but nice to have the capability for it • We never had a chance yet to test the “T” of the TDAQ (i. e. triggering in the “NA 62” way)

3. L 0 Trigger Processor

L 0 TP

My views on it • Some progress, mostly coherent, but still most critical single item: no proof of principle yet • Parts of PC-based solution tested • Alternatives: all-in-FPGA (but cannot just copy from TALK) • Need a definite plan for fall 2013 dry run testing to demonstrate feasibility

4. GPU triggering

GPU RICH L 0 trigger

GPU straws L 1 trigger

To be continued: - Add pile-up - GPU coding

GPU data transfer

My views on it It now really looks like this side-project can be implemented realistically It can be quite useful for NA 62 to allow a wider physics agenda Large interest in general, planning workshop in spring

5. Challenges & plan

Simulation Work on L 1/L 2 is still missing (due to manpower & Monte. Carlo) Implications on e. g. computing have now become serious as expected: cannot reliably estimate size of online farm or downscaling factor required: with no L 1/L 2 collected statistics might be 1. 5% of nominal Small task force really required

Next dry runs Fall 2013: Interested parties: LKr (CERN), LKr/L 0 (Roma Tor Vergata, Perugia), CEDAR, L 0 TP (Ferrara, Torino), TDAQ (Pisa), … Proposal: 17 -30 November Spring 2014 ? Interested parties: Straws Put forward your requests now “Continuous” dry run: Towards system availability & expertise 24/7 for testing