Set randy factor

From Hall A Wiki
Jump to: navigation, search

[Original entry]


We fixed this problem by increasing the event buffer size and changed the "randy_factor".

The event buffer size was set to accommodate the expected maximum size for a pedestal run:

nmods = 16

nchannels/mod = 16

nstrips/chan = 128

nsamples/strip = 6

nbytes/word = 4

  size = 16 * 16 * 128 * 6 * 4 = 768kB

To be safe, we set this value to 800kB:

   #define MAX_EVENT_LENGTH   1024*800     /* Size in Bytes */


The CODA ROC (by default) expects events that are 4MB / 64 = 64kB. 4MB is the total network buffer size, and 64 is the default "randy_factor".

The 4MB size cannot be changed.. so we change the randy_factor to allow for larger events:

randy_factor = 4MB / 800 kB = 5.12 -> 5


To set this value... we used a TCL script (setrf.tcl) loaded when the ROC is executed:

coda_roc_rc3 -type ROC -name ROCNAME -i -f setrf.tcl


Where setrf.tcl:

   proc setrf {val}
   {
       global env
       set rocname [info objects]
       $rocname configure -randy_factor $val
       set myvar [$rocname cget -randy_factor]
       puts "setrf -> $myvar"
       return $myvar
   }
   setrf 5


The routine first gets loaded by the ROC, then executed with the

   setrf 5

to set the randy_factor to 5. The ROC is run in interactive mode here... so one could also change the factor on the fly with

   setrf <value>

at the ROCs TCL shell.