Some plots of AIDA dead time (From PAUSE/RESUME data items) against time.

The histograms are binned in intervals of 100 ms, with the blue bars corresponding to dead time in the FEE - the y-value is the percent of the 100ms interval the card was dead.

Beam spills are indicated by the red bars (1 = on spill)

Dead time is mostly dominant in the n+n FEES: fee2, fee4, fee6, fee8 and a 30 second snapshot is shown
Can see deadtime largely seems to occur at the end of a spill - this suggests buffers/queues filling up from the high on-spill rates .

The much quieter p+n strips show a similar pattern but much less frequently.

Deadtime does not always sync up between FEE64s it seems.

IN all case the pause/resume interval seems to only last for less than 100 ms.