Difference between revisions of "DAQ Meeting/20180803"
From PREX Wiki
Jump to navigationJump to search| Line 15: | Line 15: | ||
==Minutes== | ==Minutes== | ||
| − | + | Thinking about Online Analysis with Ciprian and Cameron | |
| − | + | * Last known working Panguin and Pan duo | |
| − | + | ** Live in apar@adaq3~/PREX/pan/ and pan/panguin/ | |
| − | + | ** online.c and online.h is the online monitor | |
| − | + | ** int main is the main program which creates a root TApp and creates pages based on a config file given as a command line argument | |
| − | + | ** config file example: prex.cfg | |
| − | + | * Functioning Japan lives in ~apar/ciprian/japan/ and in ~a-onl/pking/japan | |
| − | + | ** ~/ciprian/japan/Parity/prminput/prexin.conf is the functioning configuration for the injector that is being used right now | |
| − | + | *** Looks at a .map file to map out the detectors in the .crl similar to old PAN control.db | |
| − | + | *** There are multiple files now that contain different segments, for modularity | |
| − | + | *** Label names in the .map file come out as the branch names in the JAPAN output | |
| − | + | ** Then you get a MPS_tree (MPS = one pulse per event, see [[DAQ_Meeting/20180524#Minutes|Meeting notes]] | |
| − | + | *** A lot of the variables currently placed into the JAPAN output files are not necessary for online analysis purposes | |
| − | + | ** PANGUIN functionality is to put the root file directly into RAM memory | |
| − | + | *** So we should use a truncated output root file to speed up online data lookup - trimmed down analysis outputs exist | |
| − | + | ** Startrun script ~/scripts/startAnalyzer_prex is required to run the online analysis (paired down version of regular analysis - runs fast and gives lots of updates) | |
| − | + | *** ~/PREX/feedback/panFFB is the fast feedback code - this script reads last N events from the ET directly (rather than the memory resident root file), so it is susceptible to hanging - this feeds data into PANGUIN for online plotting - fast event rates will be a problem and needs to be updated (wont work for MOLLER unless paralleled to death with GPUs etc.) | |
| − | + | **** These fast feedback executables will sometimes choose to block the ET whenever sequential events are unable to be read | |
| − | + | **** This is so that you can make accurate helicity correlated beam corrections, etc. and not get back results due to pileup | |
| − | + | **** We absolutely need to turn it off during production online analysis | |
| − | + | **** There is a control.db for fast feedback that you can set to change beam, etc. parameters so that you can study beam quality, etc. in your raw data output | |
| − | + | *** Prompt analysis is another analysis which is done automatically at the end of a run and reads the data written to disk (adaqfs) | |
| − | + | *** WAC analysis is a third type of analysis is some sort of corrective post-analysis done by hand by Weekly Analysis Coordinators (WAC) | |
| − | |||
== Testing == | == Testing == | ||
[[DAQ Testing/20180803|Today's testing]] | [[DAQ Testing/20180803|Today's testing]] | ||
Revision as of 13:53, 3 August 2018
Back to Main Page >> DAQ Documentation Portal >> DAQ Meetings
previous meeting << >> following meeting
Logistic information
BlueJeans calling instructions: Toll-Free Number (U.S.& Canada): 888-240-2560 International toll number: 408-740-7256 Bluejeans CODE: 475 839 391 Bluejeans link: https://bluejeans.com/475839391
Agenda
- Bob :
- Ciprian : revive panguin and other online analysis features
Minutes
Thinking about Online Analysis with Ciprian and Cameron
- Last known working Panguin and Pan duo
- Live in apar@adaq3~/PREX/pan/ and pan/panguin/
- online.c and online.h is the online monitor
- int main is the main program which creates a root TApp and creates pages based on a config file given as a command line argument
- config file example: prex.cfg
- Functioning Japan lives in ~apar/ciprian/japan/ and in ~a-onl/pking/japan
- ~/ciprian/japan/Parity/prminput/prexin.conf is the functioning configuration for the injector that is being used right now
- Looks at a .map file to map out the detectors in the .crl similar to old PAN control.db
- There are multiple files now that contain different segments, for modularity
- Label names in the .map file come out as the branch names in the JAPAN output
- Then you get a MPS_tree (MPS = one pulse per event, see Meeting notes
- A lot of the variables currently placed into the JAPAN output files are not necessary for online analysis purposes
- PANGUIN functionality is to put the root file directly into RAM memory
- So we should use a truncated output root file to speed up online data lookup - trimmed down analysis outputs exist
- Startrun script ~/scripts/startAnalyzer_prex is required to run the online analysis (paired down version of regular analysis - runs fast and gives lots of updates)
- ~/PREX/feedback/panFFB is the fast feedback code - this script reads last N events from the ET directly (rather than the memory resident root file), so it is susceptible to hanging - this feeds data into PANGUIN for online plotting - fast event rates will be a problem and needs to be updated (wont work for MOLLER unless paralleled to death with GPUs etc.)
- These fast feedback executables will sometimes choose to block the ET whenever sequential events are unable to be read
- This is so that you can make accurate helicity correlated beam corrections, etc. and not get back results due to pileup
- We absolutely need to turn it off during production online analysis
- There is a control.db for fast feedback that you can set to change beam, etc. parameters so that you can study beam quality, etc. in your raw data output
- Prompt analysis is another analysis which is done automatically at the end of a run and reads the data written to disk (adaqfs)
- WAC analysis is a third type of analysis is some sort of corrective post-analysis done by hand by Weekly Analysis Coordinators (WAC)
- ~/PREX/feedback/panFFB is the fast feedback code - this script reads last N events from the ET directly (rather than the memory resident root file), so it is susceptible to hanging - this feeds data into PANGUIN for online plotting - fast event rates will be a problem and needs to be updated (wont work for MOLLER unless paralleled to death with GPUs etc.)
- ~/ciprian/japan/Parity/prminput/prexin.conf is the functioning configuration for the injector that is being used right now
