Right now, every hardware read/write is executed as a single transaction (OK, not quite, as there is the possibility to read and write a list of supplied registers)
It needs to be known whether we suffer any meaningful amount by wrapping our IPBus transactions in this way, a la TCDS.
Where this may become an issue is when we do many things at a time, so here we may benefit from understanding better whether we take a hit wrapping each single read/write in a call that includes the dispatch call, rather than queuing all the read/writes and then executing the dispatch.
Right now, every hardware read/write is executed as a single transaction (OK, not quite, as there is the possibility to read and write a list of supplied registers)
It needs to be known whether we suffer any meaningful amount by wrapping our IPBus transactions in this way, a la TCDS.
Where this may become an issue is when we do many things at a time, so here we may benefit from understanding better whether we take a hit wrapping each single read/write in a call that includes the dispatch call, rather than queuing all the read/writes and then executing the dispatch.