DUNE FD DAQ Firmware Status (October DAQ Sprint Summary) David - - PowerPoint PPT Presentation

DUNE FD DAQ Firmware Status (October DAQ Sprint Summary)

David Cussans DUNE Upstream DAQ Meeting 29/Oct/19

Outline

• What did we try to do?
• How much did we get done?
• Outstanding items?
• What can we get done in Monday “DAQ Days”?
• How long will it take?

David Cussans | DUNE DAQ Firmware 2

Activities in DAQ Sprint?

• Integration of hit finding code with Felix:
• Build environment
• Interface to central router
• Debugging hit finding
• Be able to merge pipelines
• State save/restore for pedestal subtraction and filtering
• Be able to find hits in all time-ticks

David Cussans | Firmware Overview 3

Embarrassingly Parallel

• Ten fibres per APA @ 9.6GBit/s
• 12-bit ADC samples @ 2 MSample/s
• Each fibre 2560 wires.
• Induction and collection wires on same fibre
• Treat each fibre in parallel
• How safe is this? Is it sufficient to perform common mode noise

subtraction on a per-fibre basis?

David Cussans | Firmware Overview 4

• Firmware: What’s in it?

David Cussans | Firmware Overview 5

40 pipelines Per APA

Firmware: Receive Data

• Data on fibres has complicated encoding.
• Francesco Gonnella: Birmingham, David Cussans: Bristol, Alessandro Thea: RAL
• Now synthesizes for Felix ( chasing timing issues. )
• Needs more testing

David Cussans | Firmware Overview 6

Firmware Framework

• Dave Newbold, RAL
• 2MSample/s from ADC, but logic can run much faster
• Run processing logic ~ 200MHz
• Data arrives packets containing all channels for a single “time tick”
• Rearrange into packets with multiple time ticks for a single channel.
• Save/restore state so processing transparent to block size

David Cussans | Firmware Overview 7

Firmware: Pedestal Subtraction, Filter

• Kostas Manoulopolous: RAL
• Can now cycle through wires

David Cussans | Firmware Overview 8

Firmware: Hit Finding

• Kunal Kothekar: Bristol
• Issues located and fixed (2 days ago)

David Cussans | Firmware Overview 9

Firmware: Interface to Felix Firmware

• Simone Ponzio: CERN , Filiberto Bonini: BNL, Alessandro Thea: RAL
• Extremely active.
• Working with Alessandro Thea towards hit finding integrated with Felix
• Felix code not written as framework for user code
• AT has written IPBus / Wupper bridge
• SP has written interface from AXI4S to Central Router.
• Adding ability to filter empty packets

David Cussans | Firmware Overview 10

Firmware: Configuration, Debugging

• Dave Newbold: RAL, Simone Ponzio: CERN, Alessandro Thea: RAL
• DMN has written code to allow channel-by-channel configuration
• e.g. wire-by-wire threshold
• Needs testing and debugging
• SP, AT working on adding ability to debug hit finding pipeline over

IPBus/Wupper

• Will be needed to fix issues.

David Cussans | Firmware Overview 11

Firmware: Build Environment

• Alessandro Thea, RAL
• Challenging to build firmware:
• Traditional approach: One engineer and their Workstation
• We have distributed team
• Mix of framework , user code
• Multiple hardware platforms
• Tools developed within LHC to facilitate collaborative firmware

development

• CMS IPBus-build (ipbb)
• Felix firmware build tools not yet configured to facilitate

distributed development ( e.g. separation of framework and processing code)

David Cussans | Firmware Overview 12

Firmware - a work in progress

• Starting from standalone hit finding in ZCU102 , KU105 eval

boards

• No input from WIB (yet)
• Integrated with FLX712
• Issues with timing closure
• Debugging in progress
• Colour code:
• magenta: mgts
• blue: central router
• green: wupper
• pale: regmap
• forest: dme if/pcie
• red: toHost fifos
• yellow: ipbus infrastructure
• light orange: hitfinder data-reception
• deep orange: hitfinder - everything else

David Cussans | Firmware Overview 13

Interface to Readout Software

• Pierre Lasorak: Sussex , Phil Rodrigues: Oxford
• Assistance from Roland Sipos
• Writing ArtDAQ Board Reader
• Will read hits produced by firmware and pass on in same

format as s/ware hit finder

• Current status: Can load data into DPR buffer and produce hits
• n host.
• Still a bit "fragile"
• Will soon move onto critical path.

David Cussans | Firmware Overview 14

Other Work in Parallel Buffering

O(10s) RAM Non-volatile ( for SN burst storage)

Firmware: Lossless Compression

• Patrick Dunne, Abbey Waldron: Imperial

David Cussans | Firmware Overview 16

Firmware: Buffer Management

• Erdem Motuk: UCL

David Cussans | Firmware Overview 17

• One DRAM interface

Per APA

Firmware: Non-Volatile Buffer

• Roy Wastie: Oxford

David Cussans | Firmware Overview 18

• NVMe

Interface

Two NVMe SSD Per APA

Summary

• Have not yet demonstrated hit finding on live data at

ProtoDUNE

• Still possible before LAr emptied.
• … but running out of slack: There will be system integration

issues.

• Working on buffering in parallel.

Presenter Name | Presentation Title 19