Downlink rates

1
Downlink rates

• The next slide was presented to Jonathans flux
  review
• It was based on the UW background runs, for which
  the generated and trigger rates are available,
  and which sampled a full orbit-day at 10.
• An important parameter downlink limit is 1.2
  Mbit/sec (average).

2
Simulating rates the big picture
Incoming rate into the 6 m2 sphere is it right?
Corresponding trigger rate implies 15 deadtime
too high!
Downlink rate saturates our bandwidth allocation?
one full day
3
The downlink rates from the SLAC/Lyon 1B
background run
Note it was not a full day
If we throw out no-cal events here, rather than
later, save a factor of two
4
What is the limit? Are we in trouble?

• Depends on
• amount of event data (including noise, which for
  the tracker is probably overestimated)
• JJs skill at lossless compression
• Current status, from JJ, via Atwood
• OVERHEAD 12 bytes (fixed)  
• GEM 40 bytes (fixed) (a little work could bring
  this done to lt20 bytes, I'll do it at some time
  cause its cheap)
•  
• ACD 7 addressing bits
•            5 bit hit mask (The 5 bits are-gt Hit
  AB side. Pha Accept AB side, GEM Veto hit)
•            12 bits for each PHA side (up to 2)
•            example A tile with only one PHA
  reporting in 24 bits
•                          A tile with both
        PHA's reporting in 36 bits
•  
• CAL 32-64 bits overhead
•         4-5 bits addressing / log
•         1217 bits for the ADC value /log (both
  ends)
•           Empirically I am measuring about a
  factor of 2-2.25 compression over the data
  encoding
•           of  16 towers (32 bits 32 bits
  nlogs)
•           Example 50 logs  64 (515) 50
  bits gt 133 bytes

5
OK, these are (almost) linear in the number of
DIGIS what do we see for these events?
6
Apply the JJ analysis
Event size analysis Event size analysis Event size analysis Event size analysis Event size analysis Event size analysis Event size analysis
  All events All events NoCal NoCal NoCal
n bits total NoCal bits tota
overhead 1.0 480 480 1.0 480 480
ACD 1.8 30 54 0.8 30 24
TKR 9.8 18 176 6.7 18 121
CAL 19.0 20 380 0  20  0
1090 625

TRK is overestimate, assuming 1 TOT/hit TRK is overestimate, assuming 1 TOT/hit TRK is overestimate, assuming 1 TOT/hit TRK is overestimate, assuming 1 TOT/hit
7
The bottom line

• 1.2 Mbits/s / 1.1 kbit/event 1.1 kHz event rate

Preliminary!