Page 1 of 1

Posted: 12 Feb 2011, 18:36
by mat
hi,
i ve uploaded a file called "OptMetronom" in the upload section:

http://www.sensomusic.com/forums/upload ... tronom.pat

my idea for my performance: usine plays the background beats/atmos, in 4 instances of ni reaktor and one ni battery. i change the programs in the reaktors with a korg nanopad and a related patch, which sends prg chng to the ensembles via internal bus.

i want to start/stop usine remotely and change tempo "stepless" with the dial on my e-drum, while usine is in play. (eg. slowly from 80 bpm to 160 etc)

the problem: changing tempo while play makes usine go out of sync and not finding the exact tempo anymore, +/- 10 or more bpm. (V 5.25i and 5.5), and the 4 reactors go out of sync, playing wild, confuse beats. so i stop usine and run again from the e-drum (usine auto restart on) and it catches sync again. i could live with that, just not changin tempo while play, though i would love to do so.

second, (bigger) problem:

usine runs great, i play my drum n base beats live to it for 20min or so and am realy excited, playing fast rolls and fills :) and suddenly usine starts hanging itself in a way that the tempo goes up and down arbitrarily and usine will not respond to start/stop (or any other midi)messages from my e-drum. it will be quiet for some time and then suddenly play half a second, then quiet again and so on. nothing but restart helps.

so the question is if i did some unnessecary or stupid things in my OptMetronome patch, which is the midi clock receive patch, confusing usine?

or if i should do it any other way i did not think about?

i have 4 tracks (V5.5 free), set usine to 4 threads, 64ms processing time, cpu protect 100. cpu load is at around 20 to 25 percent (task manager) using my i5 timelineX laptop.

Posted: 12 Feb 2011, 20:01
by nay-seven
many things to check :

what about memory level..?

cpu protect 100 is not a good idea , 95% is a better one

64 ms really needed..? more often 128 is enough

what about Refresh Speed..?

have you Windows virtual memory disabled..? ( recommended )

Posted: 13 Feb 2011, 12:47
by mat
memory: i have 4 gig, about 1.5 to 2 used in the set.

cpu: the help says with multiprocessor you can go to 100. anyway, i will test it with 95. only 35 percent used max in my setup.

64: i will test 128. but for fast live drumming i thought 64 would be best.

i run the drums audio through usine for FX-processing, and battery doubles up the drum sounds (so usine handles midi clock and the drum notes on one midi input, thats not good, i know, but the drum only has one midi port). but no way to remote start/stop without sending midi clock i think.

virtual memory disabled.

refresh speed? what is that...? :)

and: i did the IRQ priority setting to max (postet here somewhere), all usb-ports are on one irq. (stupid idea of acer. 3 ports-one irq) this gave my old soundcard crackle free audio on 256 samles latency.

i was hoping i did a mistake in the patch i posted. would be the simplest solution :)

Posted: 13 Feb 2011, 13:28
by nay-seven
memory: i have 4 gig, about 1.5 to 2 used in the set.
was more talking about the memory meter in Usine
64: i will test 128. but for fast live drumming i thought 64 would be best.
sometimes best is the enemy of good.
refresh speed? what is that...? smile
global setup/asio
i was hoping i did a mistake in the patch i posted. would be the simplest solution smile
don't have check it deeply, I'm not a synchro specialist , have you try the simplest receive midi clock included in Usine..?

Posted: 13 Feb 2011, 15:16
by Floego
You could try a fast workaround (as a way to avoid restarting usine everytime the problem arises):

-Go inside the usine setup/midi_in tab.
Image
-Now temporarily set the assigned midi_in port to "none" ,
Image
-Then, assign the midi_in port again as it was set before you disabled it.

try and tell :)

Posted: 13 Feb 2011, 17:42
by mat
thanx nay-seven and floego for the tips.

i will work on that hopefuly tomorrow again and tell.

Posted: 15 Feb 2011, 13:04
by mat
it seems to work now, one hour playing and no hangup.

the key was i guess the 64ms bloc set to 128.

and maybe the (i think new) midi clock patch in 5.5, using 5 midi in modules instead of one in-module and 5 filters.

tempo realtime changing is not perfect yet, but well ... less is more.

now cpu in usine is at around 4 to 22%, which is nothing compared to the fat ensembles i use in 5 reaktors! omg....

and all this with asio refresh set to 1ms.

and with a stone old M-audio mobile pre on 256samples.

time for a gold membership.

Posted: 15 Feb 2011, 19:16
by nay-seven
cool, good news !