100 likes | 201 Views
Common DAQ Design options and CPU-based sparsification D. Haas. Outline Inputs from Strasbourg/Bonn Communication Critical points to solve State Machines/Interconnections Event processing Conclusions. Common DAQ - Inputs. Existing system from Strasbourg:
E N D
Common DAQ Design optionsand CPU-based sparsification D. Haas Outline Inputs from Strasbourg/Bonn Communication Critical points to solve State Machines/Interconnections Event processing Conclusions
Common DAQ - Inputs • Existing system from Strasbourg: • Single PC, WinXP, Borland C++ • simple writing of acquired data on disk • Existing system from Bonn/Mannheim: • Single PC, WinXP, Borland C++ • more flexible producer/writer system • Both lack communication interface to other PCs D. Haas, JRA-1 Review, 4. April 2006, Page 2
Adding communication • Use existing solutions in HEP community and adapt DAQ accordingly • XDAQ with SOAP, I2O are good candidatesBUT: XDAQ is *nix only (Linux, BSD, Mac OS) • SOAP, I2O can communicate between different platforms • SOAP is XML based, slow (Commands) • I2O fast, binary format system (Data) • Build custom solution based on simple TCP/IP socket communication D. Haas, JRA-1 Review, 4. April 2006, Page 3
Adding communication • Use maximum of existing DAQ and add only necessary communication layer and ‘central’ control • Looks feasible for the Bonn system (but needs to be checked) • Or: Use minimum of existing DAQ (HW-level, readout, eventually event-format) and add own system around • seems more realistic for Strasbourg system D. Haas, JRA-1 Review, 4. April 2006, Page 4
Present Bonn DEPFET DAQ Hardware Hardware Hardware telescope ‘producer’ task DEPFET ‘producer’ task other ‘producer’ tasks DAQ buffers Writer task file Monitoring buffers Monitoring task 1 Monitoring task 2 D. Haas, JRA-1 Review, 4. April 2006, Page 5
Proposed Bonn Future DAQ DUT DAQ PC - WIN (beam area, via VNC) main PC - WINDOWS (counting room) USB DUT1 hardware Producer TCP Mon 1 PCI DUT2 hardware Producer writer MON PC - LINUX (remote) telescope DAQ PC - LIN (beam area, via VNC) VME telescope hardware Producer Mon 2 TCP TCP CAMAC TDC hardware Producer D. Haas, JRA-1 Review, 4. April 2006, Page 6
Critical points to solve • Communication via TCP/IP: • Add separate Controller TasksMessaging via SOAPData transfer via I2O • Main PC • Implement Global Controller Task (Run Control) • Monitoring • Use rootd for Monitoring (OS independent) • Write separate monitoring buffers/files D. Haas, JRA-1 Review, 4. April 2006, Page 7
State machines/Interconnections • Currently starting to define the state machines for controller processes • Interconnections will be similar to Bonn proposal D. Haas, JRA-1 Review, 4. April 2006, Page 8
Event processing • Initially, CPU has to provide CDS and sparsification for Strasbourg board • Mixed Data Mode should be foreseen (raw data + sparsified data) for easy debugging and analysis • Sparsification should be done on chip later • CPU has to take care of clusterization and complete event building (tracking) D. Haas, JRA-1 Review, 4. April 2006, Page 9
Conclusions • Code from Strasbourg and Bonn in our hands now • Analysis of the code and design of the framework will start after this review • Emlyn Corrin will join our group in July • Global run control should be OS-independent D. Haas, JRA-1 Review, 4. April 2006, Page 10