April 15, 2020 online meeting minutes: Difference between revisions
Jump to navigation
Jump to search
Created page with "present: Sergey, Valery, Ben, Stepan, Rafo, Nathan, Gagik Valery: physics motivation for DAQ/Trigger upgrade electron trigger: rag - 30kHz, rgb- 20kHz muon trigger: biggest..." |
No edit summary |
||
| Line 1: | Line 1: | ||
present: Sergey, Valery, Ben, Stepan, Rafo, Nathan, Gagik | present: Sergey, Valery, Ben, Stepan, Rafo, Nathan, Gagik | ||
Valery | * Valery presented first draft of the physics motivation for DAQ/Trigger upgrade; as we discussed on last meeting, expected electron trigger rates for rg-a - 30kHz, rg-b - 20kHz; muon trigger biggest expected rate for inbending Rg-b 20kHz; mesonX (photo production): 10kHz; Valery will prepare more accurate number by next meeting, preliminary total expected rate is 70kHz, which will require DAQ to be able to run up to 100kHz - very much limit for L1-based system | ||
We discussed again possible L1 improvements. In particular, Q**2 cut helps but may require better DC roads resolution, Valery will prepare more clear description on existing limitations. We may be able to shrink coincidence timing windows in various places to decrease accidentals. Still DC and EC related triggers needs to be improved as we discussed before. | |||
* Ben/Sergey: VTP-based streaming front-end status was reported. Big step was made by Ben fixing TCP/IP problem in VTP firmware. After that fix DAQ was running for several days and dropped VTP links were never observed. Sergey added lifetime-related statistics to coda_sro. Sergey installed redhat devtools-8 on clonfarm1 (after permission from CC), and TriDAS team was able to extract TriDAS from GIT and compile it locally, testing is started. Ben will do more adjustment to VTP firmware to improve TCP performance, currently looks like we have a limit at 300-500MB/s, cannot tell more precisely at that moment. | |||
* Gagik gave preliminary description of his ideas how to organize online data processing in L3 and streaming framework, will prepare document by next meeting; it will be discussed more since we do not have yet clear picture how to proceed, in particular what will be the role of AI, existing offline algorithms and L1 algorithms in streaming daq | |||
* Sergey will prepare review of existing L1 algorithms and how it can be improved - by next meeting | |||
* Rafo assignments was briefly discussed. One of possible areas will be to estimate possible L1 improvements by processing existing data offline, will be discussed more after we'll set workplan. | |||
Rafo assignments | |||
Revision as of 12:09, 15 April 2020
present: Sergey, Valery, Ben, Stepan, Rafo, Nathan, Gagik
- Valery presented first draft of the physics motivation for DAQ/Trigger upgrade; as we discussed on last meeting, expected electron trigger rates for rg-a - 30kHz, rg-b - 20kHz; muon trigger biggest expected rate for inbending Rg-b 20kHz; mesonX (photo production): 10kHz; Valery will prepare more accurate number by next meeting, preliminary total expected rate is 70kHz, which will require DAQ to be able to run up to 100kHz - very much limit for L1-based system
We discussed again possible L1 improvements. In particular, Q**2 cut helps but may require better DC roads resolution, Valery will prepare more clear description on existing limitations. We may be able to shrink coincidence timing windows in various places to decrease accidentals. Still DC and EC related triggers needs to be improved as we discussed before.
- Ben/Sergey: VTP-based streaming front-end status was reported. Big step was made by Ben fixing TCP/IP problem in VTP firmware. After that fix DAQ was running for several days and dropped VTP links were never observed. Sergey added lifetime-related statistics to coda_sro. Sergey installed redhat devtools-8 on clonfarm1 (after permission from CC), and TriDAS team was able to extract TriDAS from GIT and compile it locally, testing is started. Ben will do more adjustment to VTP firmware to improve TCP performance, currently looks like we have a limit at 300-500MB/s, cannot tell more precisely at that moment.
- Gagik gave preliminary description of his ideas how to organize online data processing in L3 and streaming framework, will prepare document by next meeting; it will be discussed more since we do not have yet clear picture how to proceed, in particular what will be the role of AI, existing offline algorithms and L1 algorithms in streaming daq
- Sergey will prepare review of existing L1 algorithms and how it can be improved - by next meeting
- Rafo assignments was briefly discussed. One of possible areas will be to estimate possible L1 improvements by processing existing data offline, will be discussed more after we'll set workplan.