giuliomoro I don't know if there is something you do with buttons and MIDI that changes the CPU processing? Your patch is too complicated for me to understand it at a glance!
sorry, of course it is a mess actually 🙂
i only start writing to tables and change which tables are played back with the buttons. midi is only for note input to change playback speed and aftertouch to scroll through the table when in "scrub" mode. i am now recompiling with your new hvcc repo, and i will disconnect all midi controllers to see if it makes a difference in load. but my CPU was a steady 69 to 70 % regardless of any midi activity or button presses. that is why i was so astonished, the simple fact that a midi device and buttons are connected should not make that much of a difference.
EDIT: ahem stupid me i guess, should i be looking at the bela-audio load? that is as you suggested 26-28% i was looking at the ROOT process which takes 70%..
Every 0.5s: cat /proc/xenomai/sched/stat bela: Wed May 8 19:10:56 2019
CPU PID MSW CSW XSC PF STAT %CPU NAME
0 0 0 4353657 0 0 00018000 70.0 [ROOT]
0 1681 9 11 26 0 000600c0 0.0 granular
0 1691 2 3 4 0 000480c0 0.0 bela-midiIn_hw:0,0,0
0 1692 2 3 4 0 000480c0 0.0 bela-midiOut_hw:0,0,0
0 1693 5 761787 761794 0 00048046 28.1 bela-audio
0 0 0 2427976 0 0 00000000 0.7 [IRQ16: [timer]]
0 0 0 380980 0 0 00000000 0.7 [IRQ180: rtdm_pruss_irq_irq]
does that look reasonable? does ROOT just take all that is left to get to 100% then? (i seem to remember some discussion in the forum between you and thetechnobear where this came up as well)