Discussion of: Throttling strategies and optimization of the trigger-less streaming DAQ system in the CBM experiment
Jump to navigation
Jump to search
- David
- Problem: interaction rates fluctuated 4-5 times with beam intensity. Design back pressure to limit data loss.
- Figures need more caption. Also, the first figure referenced is fig. 9 (i.e. out of order)
- Page 2, bottom of first column: acceptable loss of 5% of hits. This seems unlikely to be distributed throughout the STS. Seems like a potential for resolution bias with rate at the very least.
- 50Mhits/s * 14,400 ASICS = 720 GHits/sec. Not clear how many bytes/hit. (1bytes/hit -> 720GB/s, 2bytes/git -> 1.44TB/s)
- delta electrons = knock-on electrons
- "stop" strategy will read out the busy events which presumably have more pileup. "clear" strategy throws away the events with higher pileup. In the end, "clear" should leave you with a cleaner data set.
- Figure 6: why is there a leading tail for "stop" but not "clear"?