As a follow up to (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=28749) we have a build of the framewriter on h1fw2 that splits the producer thread into two components.

Here is a sample timing:

producer thread timings
 recv min:0.0592079 mean:0.0624186 max:0.0642519
producer crc thread timings
 full min:0.0259149 mean:0.0270363 max:0.0351679
 crc  min:0.014359 mean:0.0144825 max:0.0150859
 xfer min:0.0115559 mean:0.0125535 max:0.0200801

The producer thread does nothing but receiving data.  Here we notice the time to receive is increased to the full 1/16s.  We believe this is largely a busy wait while it waits for the data to come in.

The secondary thread (producer crc) is now doing the crc calculations and pushing the data stream into the main buffers.  We will watch this over night and see how the system is running tomorrow.