stefan.ballmer@LIGO.ORG - posted 11:28, Monday 04 November 2013 - last comment - 09:14, Wednesday 06 November 2013(8369)
Another look at the EPICS delay test setup
Attached is a plot of the round trip delay times of EPICS PULL (active read) EPICS PUSH (active write) and EPICS PUSH/PULL (active write to 3rd channel, followed by active read from that channel). PUSH and PUSH/PULL are dominated by a 2Hz update cycle. PULL sees a 8Hz update cycle. The minimum delay for PUSH is around 110msec, for PULL and PUSH/PULL it is about twice that. The delay for all three channels is about 10x too slow to be really useful (i.e. for acknowledgment returns shorter than the blink of an eye.)
Images attached to this report
Comments related to this report
A last snapshot of the EPICS test setup. Note that the intermediate EPICS server for the Push/Pull line was rebooted. an EPICS error was indeed reported. I backed out the temporary changes for testing from h1odcmaster and h1ascimc. Both models were re-compliled and re-installed and restarted without the changes.
Images attached to this comment
