Creating additional MIDI ports causes increase in CPU load (possibly?)

Cantabile
1

Will this affect the performance for midi ports? It seems to me that the cpu usage increases if I add more ports to a rack, but not if I use bindings to channel events/notes somewhere. I might be mistaken, though.

2

Just adding MIDI ports to a rack shouldn’t affect performance. If it is, I’d be interested in seeing a Profiler log for both cases to see what’s going on.

3

I’ve send you a song with two embedded racks that demonstrates this - one rack with more ports, and the profiler tells that the rack with fewer ports uses 0,2 % in average, whereas the other uses 0,9% in average.

4

Thanks - I’ll take a look at this during the week.

1 Like
5

Thanks. If someone else want to take a look or chip in, here is the
song (635.9 KB) that demonstrates the issue for me.

Inspired by @terrybritton I created a rack with named ports, to introduce an abstraction layer on top of my SL Mixface, to make it easier to recognise the events. I made a port for each button/fader/knob, and used the rack in all my instrument and effect racks - and noticed a clear increase CPU usage in the profiler. Then I changed the rack to have fewer out-ports, just having one for buttons (with a number of combinations), and used the CC number to indicate which button, and that reduced the CPU load.

I have the two racks in this song, one - More - with a lot of ports, and the other - Less - with limited number of ports. On my setup the average load for More is 0,9%, and for Less it is 0,2%. Quite a difference, especially for an embedded rack that I use a lot.

1 Like
6

Quick question, could this be reproduced by anyone?

7

Hi @TorstenH

I did have a quick look at this, and could reproduce a slight load increase with the additional ports, but the profiler showed it spread around and not necessarily related to something specific about the port - at least nothing I could easily nail down.

Since the new audio engine is dramatically different in this area I’ve decided to leave further investigation on the old engine as I’m unlikely to make any big changes there and review what happens with the new engine - when I get it up and running.

Brad

1 Like
8

The only difference between the two racks are the number of ports, bindings and routes are identical, so I assumed the ports were the culprit, but did not investigate further.

Anyway, my question was initially targeted at the new engine, really looking forward to that upgrade.