This is a topic that will be familiar to anyone who has experience in the world of [[Virtual pipe organ|Virtual Pipe Organs]].
Latency on an organ is the time between you pressing the note and hearing it, usually measured in milliseconds (ms). Typically the faster the piece you are playing, the more latency matters - you need a responsive organ. It can be the difference between whether you can play that Toccata or not.
On a Raspberry Pi 2, with 8 organ stops pulled, typically you can expect a note to respond fully in about 10 milliseconds. Technically this means the Midi <tt>note_on</tt> message has been transmitted on all 8 Midi channels.
To relate this to a normal organ playing experience, something like a 50ms 'attack time' would normally be deemed acceptable with a tracker action, though clearly this will always be a bit subjective. (If interested [<googa>http://www.pykett.org.uk/the_physics_of_organ_actions.htm |The Physics Of Organ Actions] </googa> talks about this topic some more).
If you would like to review the latency you are experiencing in Box Of Stops, please take a look at the [[Logging output]].
One other factor in the perceived latency of playing Box Of Stops is the choice of instruments on the [[Stop|Stops]] in your [[Registration]].
Ideally you are looking for instruments with a short [<googa>https://en.wikipedia.org/wiki/Synthesizer#Attack_Decay_Sustain_Release_.28ADSR.29_envelope |attack phase] </googa> - how long the sound takes to go from zero to its peak level.
A classic example of an instrument with a slow attack phase is the Pan Pipes, with the lovely breathy start to each note. However mixed in with other instruments with a quicker attack phase, the overall effect will sound quite 'raggedy'.