DAQ coordinator:
David Emschermann


The CBM data challenge


  • fixed target setup to investigate the QGP phase diagram in region of high baryon-densities
  • very high interaction rate environment: 105 - 107/s (A+A), up to 109/s (p+A)
  • fast and radiation hard detectors with free-streaming readout electronics
  • high-speed Data AQuisition (DAQ) system
  • FPGA based readout chains complemented by  state of the art computing infrastructure allowing for online event reconstruction
  • more than 5.000 GBT links operating at 4.8 Gbps as data source
  • about 2 TB/s bandwidth to the Green Cube

CBM readout in a nutshell

  • Collect raw data from subsystems
  • Pre-process data in FPGAs
  • Send micro-slices to Green Cube
  • Pack data into time-slices
  • Deliver time-slices to online analysis
  • Digest a total bandwidth of 2 TByte/s

CBM readout topology with CRI


HTG-Z920 - Common Readout Interface (CRI) prototype country of origin: California

  • Xilinx ZYNQ UltraScale+ FPGA - next generation board
  • will be operated in the FLES input node
  • to cover the functionality of both the AFCK and FLIB in a single FPGA board


ASUS ESC8000 G3 Server (FLES Input Node) country of origin: Taiwan 

  • FLES input node
  • can take up to 8x FLIB or 4x CRI
  • used to receive data in micro-slices 
  • forwards data to the FLES compute nodes


Mellanox SB7800 – EDR 100 Gb/s InfiniBand Smart Switch country of origin: Israel

  • interface between FLES input and FLES compute stage
  • will allow to transfer up to 1.2 Tb/s over 96 optical fibers

mCBM preparation

Optical fiber connection between the mCBM cave, DAQ container and  the Green IT Cube


mCBM readout scheme for 2019 DAQ3

Key Project Institutes:

  • GSI, Darmstadt, Germany
  • CQU, Chongqing, China
  • FIAS, Frankfurt, Germany
  • IRI, Frankfurt, Germany
  • KIT, Karlsruhe, Germany
  • USTC, Hefei, China
  • VECC, Kolkatta, India
  • WUT, Warsaw, Poland