L1Calo Phase-1 architechure

Slides:



Advertisements
Similar presentations
JFEX Uli Schäfer 1 Uli Intro / overview / issues Draft, will change !!!! Some questions flagged.
Advertisements

The First-Level Trigger of ATLAS Johannes Haller (CERN) on behalf of the ATLAS First-Level Trigger Groups International Europhysics Conference on High.
Phase-0 Topological Processor Uli Schäfer Johannes Gutenberg-Universität Mainz Uli Schäfer 1.
L1Calo – towards phase II Mainz upgraders : B.Bauss, V.Büscher, R.Degele, A.Ebling, W.Ji, C.Meyer, S.Moritz, U.Schäfer, C.Schröder, E.Simioni, S.Tapprogge.
ATLAS L1 Calorimeter Trigger Upgrade - Uli Schäfer, MZ -
Phase-0 topological processor Uli Schäfer Johannes Gutenberg-Universität Mainz Uli Schäfer 1.
Level-1 Topology Processor for Phase 0/1 - Hardware Studies and Plans - Uli Schäfer Johannes Gutenberg-Universität Mainz Uli Schäfer 1.
Uli Schäfer 1 S-L1Calo upstream links architecture -- interfaces -- technology.
Uli Schäfer 1 (Not just) Backplane transmission options.
S. Silverstein For ATLAS TDAQ Level-1 Trigger updates for Phase 1.
Samuel Silverstein Stockholm University L1Calo upgrade hardware planning + Overview of current concept + Recent work, results.
Uli Schäfer 1 (Not just) Backplane transmission options Uli, Sam, Yuri.
S. Silverstein For ATLAS TDAQ Level-1 Trigger upgrade for Phase 1.
Uli Schäfer 1 JEM1: Status and plans power Jet Sum R S T U VME CC RM ACE CAN Flash TTC JEM1.0 status JEM1.1 Plans.
Uli Schäfer 1 (Not just) Backplane transmission options.
Samuel Silverstein Stockholm University L1Calo upgrade discussion Overview Issues  Latency  Rates  Schedule Proposed upgrade strategy R&D.
JFEX Uli Schäfer 1 Mainz. Jet processing Phase-0 jet system consisting of Pre-Processor Analogue signal conditioning Digitization Digital signal processing.
Status and planning of the CMX Philippe Laurens for the MSU group Level-1 Calorimeter Trigger General Meeting, CERN May 24, 2012.
Uli Intro / overview / issues
JEP HW status and FW integration plans Uli Schaefer and Pawel Plucinski Johannes-Gutenberg Universitaet Mainz Stockholm University.
CMX (Common Merger eXtension module) Y. Ermoline for CMX collaboration Preliminary Design Review, Stockholm, 29 June 2011.
Topology System Uli Schäfer 1 B.Bauß, V.Büscher, W.Ji, U.Schäfer, A.Reiß, E.Simioni, S.Tapprogge, V.Wenzel.
Atlas L1Calo CMX Card CMX is upgrade of CMM with higher capacity 1)Inputs from JEM or CPM modules – 40 → 160Mbps (400 signals) 2)Crate CMX to System CMX.
JFEX baseline Uli Schäfer 1 Uli. Intro: L1Calo Phase-1 System / Jets Uli Schäfer 2 CPM JEM CMX Hub L1Topo ROD JMM PPR From Digital Processing System CPM.
CMX status and plans Yuri Ermoline for the MSU group Level-1 Calorimeter Trigger Joint Meeting CERN, October 2012,
CMX status Yuri Ermoline for the MSU group Mini-TDAQ week, CERN, 9-11 July 2012,
Status and planning of the CMX Wojtek Fedorko for the MSU group TDAQ Week, CERN April , 2012.
CMX Hardware Status Chip Brock, Dan Edmunds, Philippe Yuri Ermoline, Duc Bao Wojciech UBC Michigan State University 25-Oct-2013.
S. Rave, U. Schäfer For L1Calo Mainz
CMX Hardware Overview Chip Brock, Dan Edmunds, Philippe Yuri Wojciech Michigan State University 12-May-2014.
Ideas about Tests and Sequencing C.N.P.Gee Rutherford Appleton Laboratory 3rd March 2001.
Uli Schäfer 1 From L1Calo to S-L1Calo algorithms – architecture - technologies.
ATLAS Trigger / current L1Calo Uli Schäfer 1 Jet/Energy module calo µ CTP L1.
Algorithms and TP Y. Ermoline et al. Level-1 Calorimeter Trigger Joint Meeting, Heidelberg, January 11-13, 2010.
CMX Hardware Overview Chip Brock, Dan Edmunds, Philippe Yuri Wojciech Michigan State University 19-May-2014.
JFEX Uli Schäfer 1 Mainz. L1Calo Phase-1 System Uli Schäfer 2 CPM JEM CMX Hub L1Topo ROD JMM PPR From Digital Processing System CPM JEM CMX Hub L1Topo.
Samuel Silverstein Stockholm University CMM++ firmware development Backplane formats (update) CMM++ firmware.
CMX Collection file for current diagrams 30-Apr-2014.
17 December, 2010 ATLAS L1Calo upgrade meeting Meeting overview Recent hardware developments, ideas.
Upgrading the ATLAS Level-1 Calorimeter Trigger using Topological Information Yuri ERMOLINE, Michigan State University TWEPP 2010, Aachen, Germany,
JFEX Uli Schäfer 1 Mainz. L1Calo Phase-1 System Uli Schäfer 2 CPM JEM CMX Hub L1Topo ROD JMM PPR From Digital Processing System CPM JEM CMX Hub L1Topo.
Samuel Silverstein, SYSF ATLAS calorimeter trigger upgrade work Overview Upgrade to PreProcessor MCM Topological trigger.
CMX: Update on status and planning Yuri Ermoline, Wojciech Dan Edmunds, Philippe Laurens, Chip Michigan State University 7-Mar-2012.
Uli Schäfer 1 Mainz R&D activities. Uli Schäfer 2 MZ R&D BLT has been built and tested (backplane transmission only). A few minor issues were found. Possible.
Introduction to L1Calo Upgrade L1Calo Collaboration Meeting Cambridge 23-Mar-2011 Norman Gee.
Samuel Silverstein, Stockholm University For the ATLAS TDAQ collaboration The Digital Algorithm Processors for the ATLAS Level-1 Calorimeter Trigger.
E. Hazen1 MicroTCA for HCAL and CMS Review / Status E. Hazen - Boston University for the CMS Collaboration.
E. Hazen -- CMS Week HCAL Back-End MicroTCA Upgrade Status E. Hazen Boston University.
EPS HEP 2007 Manchester -- Thilo Pauly July The ATLAS Level-1 Trigger Overview and Status Report including Cosmic-Ray Commissioning Thilo.
L1Calo Upgrade Phase 2 ● Phase 2 Functional Blocks? – Thoughts on L1 “refinement” of L0 ● Simulation framework ● Phase 1 Online SW Murrough Landon 2 February.
Phase2 Level-0 Calo Trigger ● Phase 2 Overview: L0 and L1 ● L0Calo Functionality ● Interfaces to calo RODs ● Interfaces to L0Topo Murrough Landon 27 June.
Phase 2 L1Calo/Calo Interface ● Introduction ● Calo RODs ● Granularity and algorithms ● Calo-L1Calo links ● L1Calo design? ● Summary Murrough Landon 9.
DAQ and TTC Integration For MicroTCA in CMS
ATLAS calorimeter and topological trigger upgrades for Phase 1
L1Calo Upgrade Phase 2 Introduction and timescales
Demonstrator Slice Possibilities and Timetable
L1Calo upgrade discussion
ATLAS L1Calo Phase2 Upgrade
MicroTCA Common Platform For CMS Working Group
Cabling Lengths and Timing
DAQ Interface for uTCA E. Hazen - Boston University
Possibilities for CPM firmware upgrade
Run-2  Phase-1  Phase-2 Uli / Mainz
The First-Level Trigger of ATLAS
Level-1 Calo Monitoring
ATLAS: Level-1 Calorimeter Trigger
CPM plans: the short, the medium and the long
CMX Status and News - post PRR -
(Not just) Backplane transmission options
FED Design and EMU-to-DAQ Test
Presentation transcript:

L1Calo Phase-1 architechure Phase-1 Overview CTP upgrade status Include muon ROIs? CMM++ design Real-time data path + readout TP design Infrastructure + RTDP + readout L1Calo - CTP interface

Phase-1 upgrade Readout to ROIB/DAQ E/g t/had clusters (CP) Pre- (Muon ROIs?) E/g t/had clusters (CP) 0.1 x 0.1 Global Merger To CTP Pre- Processor (PPr) Analog tower sums (0.1 x 0.1) Cluster ROIs Jet / SET (JEP) 0.2 x 0.2 Jet ROIs High-speed fiber links

CTP upgrade studies More algorithms require more CTP inputs CTP can modify firmware to run at higher speed, multiplex inputs to accommodate more bits Possibility to send muon ROIs from MUCTPI to our topological processor Described in TDAQ week talk by Stefan Haas

CTP upgrade status (S. Haas) Modified firmware tested successfully on CTP reference system Clock phase scan shows good timing margins Valid data window 65-70% (8-9 ns) of the bit period (12.5 ns) Costs: reduced flexibility, 2-3 extra BCs of latency 124 Trigger inputs

MUCTPI upgrade (S. Haas) MUCPTI receives muon candidates from 208 trigger sectors Up to 2 muon candidates/sector/BC Muon candidate consists of pT and location information (RoI) MUCTPI now sends multiplicity per pT threshold to CTP Detailed muon ROIs sent to LVL2 and DAQ at L1A rate Topological trigger could potentially profit from muon RoI information

MUCTPI (S. Haas) CTP may have to modify MIOCTs for additional RPC chambers in the ATLAS feet region Requires 14 inputs per octant Might allow opportunity to add muon ROIs to topological trigger! We need to keep open the possibility for accepting/using muon ROIs in the TP

CMM++ design concept Collect ROI coordinates Xilinx Virtex 6 Glink DAQ/ROI readout on legacy RODs Glink Collect ROI coordinates from processor modules via backplane (4x speed) SNAP12 Up to three 12-fiber bundles 6.4 Gbit/s/fiber (can mount either Tx or Rx) Xilinx Virtex 6 XCE6VLX500T FF1759 840 I/O, 36 GTX transceivers SNAP12 SNAP12 LVDS cable outputs to CTP (legacy interface)

Staging: "Day-1" running X 2 Fan-in fiber patch cable CP 0 Jet 0 CP 1 SNAP12 CP 0 SNAP12 Jet 0 4 SNAP12 CP 1 12 40 Mbit/s 4 40 Mbit/s SNAP12 CP 2 4 SNAP12 Jet 1 SNAP12 CP 3 40 Mbit/s 12 To CTP To CTP

Stage 2: ROIs from CPM/JEP SNAP12 CP 0 SNAP12 Jet 0 4 SNAP12 CP 1 12 160 Mbit/s 4 160 Mbit/s SNAP12 CP 2 4 SNAP12 Jet 1 SNAP12 CP 3 160 Mbit/s 12 Day-1 algorithms on full ROI map To CTP To CTP

Stage 2a (optional) : Topological triggers using CMM++ SNAP12 CP 0 SNAP12 Jet 0 4 SNAP12 CP 1 12 4 160 Mbit/s 4 160 Mbit/s SNAP12 CP 2 4 4 4 SNAP12 Jet 1 SNAP12 CP 3 160 Mbit/s 12 12 To CTP To CTP JEP1 sees reduced ROI map from CP

Stage 3: Parasitic TP running X 2 SNAP12 CP 0 SNAP12 Jet 0 4 SNAP12 CP 1 12 160 Mbit/s 4 To TP 160 Mbit/s SNAP12 CP 2 4 SNAP12 Jet 1 To TP SNAP12 CP 3 To TP 160 Mbit/s 12 To TP To CTP To CTP Continue Day-1 algorithms

Stage 4: full TP running X 2 CP 0 Jet 0 CP 1 160 Mbit/s To TP SNAP12 CP 0 SNAP12 Jet 0 SNAP12 CP 1 160 Mbit/s To TP 160 Mbit/s SNAP12 CP 2 SNAP12 Jet 1 To TP To TP SNAP12 CP 3 160 Mbit/s To TP Duplicate cables to multiple TP modules

CMM++ readout Virtex 6 Original idea: 2 Glinks Either original components, or emulate in FPGA Use existing RODs ROI readout ~ same length But DAQ links need to read out 4x as much input data 35  ~97 ticks/time slice Does this limit our readout capacity? To increase r/o rate to DAQ: Use 2 or more Glink outputs for DAQ r/o Build more RODs? Glink Glink SNAP12 Virtex 6 SNAP12 SNAP12 13 13

RODs for CMM++ readout Option 1: use one of our spare RODs to read out 10 extra CMM++ Glinks Spare situation? Option 2: Make more copies of current ROD Easy to integrate Old components hard to find/replace Option 3: new RODs? Similar architecture but newer components (easier to port firmware/software) 14 14

Topological processor module Optical links Quadrant merging Optical links bring L1 ROIs data to each TPM Data preprocessed at quadrant level. Global processing of selected results Scalable: upstream replication of input signals multiple modules can run in parallel each TPM has access to full data. OPTO Control Monitoring TTC DCS DAQ Interfaces FPGA OPTO OPTO FPGA OPTO OPTO FPGA (Global) To CTP OPTO OPTO FPGA OPTO DAQ Readout interface OPTO OPTO FPGA ROI OPTO OPTO 15

Bringing muon ROIs to TP MUCTPI has 16 MIOCTs Each with ca 4  12 ROI positions (TGC coverage < 2.4) 6 thresholds per ROI (encode in 3b)  144 bits/MIOCT/BC 6.4 Gb/s fiber links 128 data bits/fiber/BC Need 2 fibers / MIOCT to each TP module So one TP quadrant would receive 40 fibers 12 EM 12 Tau 8 Jet (Each JEP crates covers 2 quadrants per JEP) 8 muon (4 MIOCTs per quadrant)

Fitting into Virtex-6 FPGAs? Virtex-6 LXT (Began shipping in March 2009) Up to 36 GTX transceivers @ 6.4 Gbit/s each Cannot bring 40 fibers to a single FPGA which we would prefer for quadrant level merging Virtex-6 HXT (samples just shipping now) Up to 48 GTX transceivers @ 6.4 Gbit/s each Up to 24 HTX tranceivers @ >11 Gbit/s each HX565T has 40 GTX transceivers But....expensive and new (long lead time?) 17

LXT vs HXT solutions LXT only LXT + HXT 18 18

TP module readout Input data: Output data: ~200 bits/BC? 1536 bits / BC / CMM++  10 144 bits / BC / MIOCT  16 Total: 17,664 bits/BC Output data: ~200 bits/BC? Readout must handle ~18,000 bits/BC! Equivalent of more than 20 G-links per TP module! We almost certainly need a new kind of ROD 19 19 19

TP ROD Option 1: TP module sends readout to external ROD crate Use multi-Gbit optical links from TP module to ROD but FPGA transceivers already needed as inputs! Option 2: TP ROD on rear transition module Option 3: TP module is its own ROD Note: rear transition module probably needed for Slink and CTP outputs, anyway (front panel full!) 20 20 20

TP and rear transition modules FPGAs 3 CTP output ROD Readout links Input links 2 1 backplane zones

Other TP infrastructure Crate/backplane/power Crate controller "TCM-like" interface TTC distribution CAN/DCS interface 22 22 22

L1Calo - CTP interface Current interface: Transitional interface: 6 direct LVDS cables from CMMs to CTP Transitional interface: 6 outputs from CMM++ plus several output cables from n TP modules Final phase-1 interface: Some number of output cables from some number of TP modules In other words, the L1Calo-CTP cabling will change over time! Option 1: Re-cable each time we change things Option 2: Build "active patch panel" to remap trigger bits to same set of CTP input cables 23 23 23

Active patch panel (conceptual) From CMM++ modules Ethernet TTC DCS/Can Router FPGA To CTP readout link(s) From TP modules

Active patch panel functions Basic function: Remap L1Calo outputs and send to CTP inputs Diagnostic functions: Capture, read out and histogram all L1Calo output bits, even those not yet sent to the CTP (useful for parasitic testing of new triggers) Custom, flexible rate monitoring Augment/extend CTP capabilities? Offload some L1Calo/muon trigger processing to the active patch panel's FPGA if needed. 25 25 25

Summary: Phase-1 hardware work CMM++ 9U ROD for CMM++ readout Topological processor TP module Crate infrastructure Controller TP-TCM TP-ROD CTP interface (active patch panel?) Plus... Test infrastructure (DSS) New fiber cable plant 26 26 26

Plus lots of firmware! JEM/CPM (add ROIs to RTDP) CMM++ CMM++ ROD Day-1, Phase 1-4, etc. CMM++ ROD Either updated 9U ROD or new module Topological processor module TPM ROD CTP interface Anything I've missed? 27 27 27